That mess burns cash through scrap, rework, and bad calls made on half the story.

I see the fix like this: start with clean, heat-linked batch data, then use machine learning where it fits. GoSmarter, built by Nightingale HQ, reads mill cert PDFs with AI OCR, splits multi-heat records, and gives metals manufacturers data they can train on without feeding rubbish into the model.

What you get from this:

• Why steel traceability matters more than fancy modelling
• Which fields matter first, like heat number, chemistry, CEQ, and Rp0.2
• Where image models fit, and where tabular models make more sense
• How to start with one defect use case and avoid alert spam
• What turns a risk score into something an engineer can act on

If you want earlier defect warnings, you need to sort the paperwork first. Here’s how to fix it.

Can Machine Learning Predict Casting Defects Before They Happen?

Why metallurgical defects are hard to catch early

Most defects show up at final inspection. By then, the scrap and rework bill is already baked in. Scrap steel usually gives you back only 40p in the pound. That means a 60% loss on every kilogram of bad material [6].

The hard part is this: defects rarely come from one bad reading. They come from process conditions stacking up in the wrong way. Weldability is a good example. It depends on Carbon Equivalence (CEQ), which you calculate from several chemical elements [3]. Miss one value, or tie it to the wrong heat, and your quality call starts from bad data. Then the mess spreads.

Your data is scattered, messy and sometimes wrong

Quality data turns up in clashing formats. Batch history ends up split across systems. Labels don’t match. Then multi-heat certificates make it worse. One document can cover several production heats, each with different chemical and mechanical properties. Generic tools often squash that into one record, or skip later heats altogether [2]. Feed that into a defect model and you train it on a warped average that matches no real batch.

Terms like Rp0.2 and +N (normalised) have exact technical meanings. Generic optical character recognition (OCR) tools often read them like plain text. That leads to mislabelled fields in quality databases [2].

The same defect rarely looks the same twice

One defect does not always wear the same face. It shifts across alloys, batches and process settings. Fixed rules struggle with that. Bad labels make it worse. The model ends up learning noise, not signal.

The defects that matter most are usually the rarest

The defects that hit hardest are usually the ones you catch too late. Manual planning often leaves scrap rates at 5-8%. Better-run systems can cut that to under 2.5% [6]. But early prediction only works when you give the model batch-level records it can trust.

Once you sort that out, you can start picking models that spot real risk instead of chasing process noise.

How machine learning turns noisy process data into defect warnings

Once you have clean batch-level records, pick the model that fits the signal. Don’t force one tool onto every problem. Structured process data needs one approach. Image-based defects need another.

Models that work well on production data

For structured process data, start with a tabular model. It’s the sensible place to begin when your data sits in rows and columns, not buried in photos or scans.

Use deep learning when the defect shows up in images

When the defect signal shows up in images, deep learning fits better. That’s the right tool for visual patterns that standard tabular models will miss. The catch is simple: these models are harder to explain. So if the model looks unsure, send those low-confidence cases for manual review [2].

Combine plant knowledge with AI to cut false alarms

AI on its own isn’t enough. Plant knowledge still does the heavy lifting. You need it to read terms like Rp0.2 and CEQ the right way [2][3]. Then you route low-confidence cases for review before they hit the quality system [2]. That keeps engineers in the loop and cuts false alarms.

What you need in place before the model is worth trusting

Defect prediction usually falls over because your batch data is a mess, not because the model is poor. Fixing that is dull work. It is also the bit that matters. Start with the heat link. If that link is wrong, the rest is guesswork.

Capture only signals you can tie to a heat

Your minimum dataset should start with signals you can tie to one heat with confidence: heat number, material grade, and chemical composition.

Heat number is the key link in your dataset [5]. If you lose that link, you can easily match records to the wrong defect outcome from the wrong batch. That is how people end up trusting a tidy chart built on bad joins.

Skip EN 10204 Type 2.1 and 2.2 as your starting point. They do not link results cleanly to a single batch. Type 3.1 or 3.2 certificate data is the safer option because it links one heat number to measured chemical and mechanical properties [3][5].

Clean the labels and select the features that actually matter

Once the batch link is clean, sort out your defect labels. Give each defect type one name. If a record looks vague or messy, pull it out for review. Do not shove rubbish into training data and hope the model sorts it out for you.

Start with fields that have clear metallurgical meaning:

• heat number
• grade
• chemistry
• Rp0.2
• CEQ [2][3]

GoSmarter, built by Nightingale HQ, includes MillCert Reader. It pulls structured data from PDF mill certificates, reads terms like Rp0.2 and CEQ properly, and splits multi-heat certificates into separate batch records [2][3]. That matters because old certificate files often look like they were designed to annoy everyone on purpose.

Test it as if production depends on it, because it does

Then test it against real production history. A model can look good in testing and still fall apart on the shop floor. Track precision and recall. Keep your validation tied to the heat number [3][5].

If the model is not sure, send that prediction to a person before you write it into the production record [2]. That step stops bad calls from slipping straight into live work.

How to get value on the line without overloading the team with alerts

A model sitting in a notebook does nothing for the line. You get value when it flags a heat, coil or batch before you create scrap, rework or a customer escape. Put that signal in front of the engineer who can act on it. Once you have clean, heat-linked data, the next job is simple: get the alert to the right person at the right time. Then you decide how live it needs to be.

Pick the setup that matches how your plant actually runs

Not every plant needs real-time alerting tied into every sensor on the line. That’s how you end up with fancy software and a team that ignores it. The right setup depends on where quality losses happen and how much extra moving parts your operation can handle without turning daily work into a slog.

Start with batch prediction if you’re still cleaning up the data. It carries less risk and shows where defects cluster across heats and grades before you wire anything into live production. Move to inline monitoring when your records are solid enough for near-live checks. Real-time alerting makes sense on high-volume lines, but only when clean data feeds it. The faster the alert, the more context it needs.

Make the output useful to engineers, not just data teams

A bare risk score is useless on the shop floor. Engineers need the reason. Chemistry out of range. Rp0.2 near the limit. A certificate that does not match the order.

The output needs to point at a lever the operator can pull - temperature, feed rate, material mix, or a hold for re-inspection - not just signal that something might be wrong.

Explainability is not a nice extra. It decides whether the model changes behaviour or gets switched off after three weeks. You need root-cause cues tied to specific metallurgical fields. That’s what makes the output believable to a metallurgist or quality engineer who has spent twenty years dealing with bad data and bad software.

Start with one record flow and prove the result fast

Start with one product family, not the whole plant. Pick something like structural hollow sections or rebar in one diameter range. Link certificates, inspection results and scrap records. That gives you enough to build a labelled dataset, train a simple model and test it against production history. Use the pilot to set thresholds, ownership and escalation rules.

Midland Steel, a major UK and Ireland rebar supplier, cut production scrap rates by 50% during real-world trials after implementing GoSmarter, built by Nightingale HQ, with AI-driven cutting plans and digital tracking; the project was led by CEO Tony Woods [4][1]. That sort of result does not come from a model running on its own. It comes from clean data, a tight starting point and a team that can see what the model is saying and why.

Conclusion: Build one clean dataset, prove one defect use case, then scale

Start small. Pick one defect, one product family, and one clean data flow.

Build the minimum heat-linked dataset first: heat number, chemistry, key mechanical properties, and heat-treatment history. Link every mill test certificate to its heat number and goods-in record, so traceability sits inside the day-to-day workflow, not buried in a PDF graveyard [1][5].

Once you clean that data, prove one defect use case. Keep it tight. You are not trying to boil the ocean. You are trying to show that clean, linked data can help you spot a problem before it burns time, scrap, and margin.

When the pilot works, roll the same setup into nearby long products. Same logic. Same data discipline. Less reinvention.

The next step is to digitise certificates into a clean, heat-linked dataset.

Next step: run GoSmarter MillCert Reader on your next batch of certificates and build one clean defect dataset

Run GoSmarter MillCert Reader, built by Nightingale HQ, on your next batch of certificates. Pull the data straight out of the PDFs instead of making someone type it all in by hand. The tool extracts heat number, chemistry, CEQ, Rp0.2, tensile strength, elongation, and heat-treatment state from each certificate. It then links each certificate back to the batch record, so you can track where the data came from [2][3][4].

Once you digitise the certificates, turn them into a clean training set. That’s the bit that matters. If your source data is a mess, your defect model will be a mess too.

These fields matter first:

GoSmarter includes automated mill cert reading, data validation checks, and a review step for low-confidence extractions before they enter the production record [1][4]. That review step matters. It stops dodgy reads from slipping into your dataset and poisoning the model.

Start small:

• Extract one batch
• Check the fields
• Fix any bad reads
• Train your first model on clean records

That’s a far better starting point than feeding junk in and hoping the software sorts itself out.

FAQs

Microsoft’s 2024 Work Trend Index found that 78% of AI users already bring their own AI tools to work, often with no guidance from their employer on how to use them well. That’s not a training problem. That’s a documentation problem. One shift lead gets a brilliant production summary out of an AI agent. The next shift lead, using the exact same tool, gets rubbish, because nobody wrote down how the first person actually asked for it.

A skills file fixes that. Instead of everyone typing a slightly different prompt and getting a slightly different result, colleagues reuse the same skill and get a consistent, useful first draft every time.

Here’s what this piece covers:

• What actually belongs in a skills file, and what doesn’t
• Why ad hoc prompting caps out at one person’s skill level
• How one branding skill file can serve marketing, sales, and technical writing at once
• A three-step framework for building your first skill this week
• Seven ready-made use cases across operations, maintenance, quality, technical, supply chain, compliance, and Learning & Development (L&D)

Let’s start with what a skills file actually is.

What a Skills File Actually Is

A skills file is not a clever one-line prompt. It’s a short document, usually less than a page, that turns one person’s good instincts into something the whole team can run. Six parts make up a solid one:

• Purpose: what the task is for, written the way a shift lead would actually say it, not a mission statement
• Inputs: the exact source material the task uses today: shift notes, a delay log, work orders, fault history, whatever it already is
• Step-by-step instructions: precisely what to do with those inputs, in order. Not “review the data.” The actual method: what to group, what to compare, what to flag
• Checks: what to verify before handing over an answer, such as every open action having an owner or every batch having a matching heat number
• Output format: the shape of the result: a short briefing, a table, an email draft, whatever the team already reads and acts on
• Escalation: the exact point where a human has to sign off, or where the AI should say “check with someone” instead of guessing

Write those six things down once, and a production manager gets the same standard of daily issue summary whether they’re pulling shift notes on a Tuesday or covering for someone on annual leave on a Friday.

GoSmarter, built by Nightingale HQ, already sits on top of a metals manufacturer’s production, quality, maintenance, and supply chain data. A skills file is the layer that tells an AI agent exactly what to do with that data, the same way, every time, no matter who’s asking.

Why Ad Hoc Prompting Doesn’t Scale Past One Person

There’s a difference between using AI and scaling AI. Using AI is one person, one chat window, one prompt they half-remember from last week. Scaling AI is a documented process that anyone on the team can pick up and run properly on day one.

Ad hoc prompting doesn’t scale because quality depends entirely on who happens to be typing that day. Your best engineer might get a genuinely useful summary out of an AI agent because they’ve spent weeks learning how to ask well. A colleague covering their shift, using the same tool, gets something vague or wrong, and either acts on it or gives up on the tool altogether.

A skills file turns one person’s good idea into a repeatable process the whole team owns. That’s the actual difference between a department that “tried AI” and one that’s scaling it.

The Branding Skill Nobody Else Should Have to Rebuild

Once departments start building skills files, the real value shows up when you stop keeping them to yourself. A skill built for one job, in one department, is often exactly what another team needs, on the same AI agent, for a completely different task.

Take branding. Marketing builds a Branding skill covering tone of voice, approved terminology, and logo and colour rules. That skill doesn’t need to stay in marketing. Sales can use the same skill to draft a customer email that actually sounds like the company. Technical can pull the same skill when writing up a case study. Nobody’s reinventing the branding rules from memory, guessing at the tone, or waiting for marketing to proofread every single document.

That’s the case for a shared skills repository: a single, central library where every department’s skills files live, instead of one shift’s good idea staying locked in one person’s head or one department’s folder. A skill built once, in one place, gets reused everywhere it fits. Good ideas stop being siloed by shift, by department, or by whoever happened to figure it out first.

Run a Departmental AI Skills Lab This Week

You don’t need a project team or a six-month rollout plan to start. You need one task, one afternoon, and three steps.

Step 1: Pick One Repeatable Task

Look for a workflow where better questions, asked the same way every time, would help colleagues get a reliable first draft. The best candidates are jobs people already spend real time on: gathering information from several sources, turning messy source material into a clean first draft, or checking something for consistency before it goes further.

Don’t pick something rare or one-off. Pick something that happens every shift, every week, or every audit cycle. That’s where a skill pays for itself fastest.

Step 2: Build the Reusable Skill

Write the six parts from earlier: purpose, inputs, step-by-step instructions, checks, output format, and escalation points. Keep it short enough that a new starter could follow it without asking a single question.

This is the step people skip. They stop at “purpose and inputs” and call it done. A skill without clear checks and a clear escalation point isn’t a skill, it’s just a slightly tidier prompt. The checks are what make the output trustworthy. The escalation point is what stops the AI guessing on something a human needs to decide.

Step 3: Raise the Use Case

A draft skill isn’t finished until someone else can use it. Turn it into a shared idea for the department: show colleagues what it does, what it needs, and what it produces. Where it fits, share it with other teams too, the way the branding skill above moved from marketing to sales to technical.

The hallmark of a strong use case is specific: named inputs, a named output format, and a named human who reviews it before it goes anywhere. “Use AI more” is not a use case. “Feed the shift notes and delay log into this skill, get a briefing with blockers and next actions, production manager signs it off before the morning meeting” is a use case.

Seven Skills Files Worth Building Right Now

Every department that touches production, quality, maintenance, supply chain, or compliance data already has a task that fits this pattern. Here are seven to start from, built around the tasks these teams already do by hand.

None of these need a data science team. Each one starts as a single department’s afternoon project, gets tested on a real week’s data, and either earns its place in the shared repository or gets binned.

Frequently Asked Questions

Build Your First Skills File This Week

Pick one task from the table above, or one of your own that fits the pattern: information gathering, turning messy source material into a clean draft, or a consistency check. Write the six parts down. Test it on a real week’s data before you show anyone.

Wales and Ireland share far more than a stretch of sea. They have long-standing trade links, a collaborative business outlook, and increasingly, a shared customer base for Nightingale HQ.

Our Chief Executive Officer, Ruth Kearney, recently met with John O’Loughlin, Head of Trade and Investment for Welsh Government in Ireland, and Iain Quick, Head of Welsh Government in Ireland, to discuss opportunities for growing Welsh innovation across the Irish market.

As a Welsh technology company, Nightingale HQ has always built products with international customers in mind. While our roots are firmly in Wales, GoSmarter.ai was designed to help manufacturers around the world improve productivity, traceability, and operational performance. Meeting the team responsible for strengthening economic ties between Wales and Ireland was a valuable opportunity as we continue to expand.

Why this meeting mattered

Nightingale HQ is entering an exciting stage of growth, with Ireland playing a key role in our expansion strategy. GoSmarter.ai, our artificial intelligence (AI) platform for metals manufacturers, is already supporting customers in Ireland, and we continue to see strong demand from manufacturers looking to automate manual processes, improve traceability, and make better operational decisions.

Welsh Government in Ireland plays an important role in helping Welsh businesses export, grow internationally, and build relationships across the Irish market. At the same time, it supports inward investment into Wales, creating opportunities that benefit businesses on both sides of the Irish Sea.

The team’s priority sectors include:

• Technology
• Construction and infrastructure
• Agritech
• Agrifood
• Offshore wind

These sectors closely align with the industries GoSmarter serves. Construction, infrastructure, and renewable energy all rely on resilient steel and metals supply chains, making Wales and Ireland natural partners for innovation in manufacturing.

“Steph and I have always built this business with a global mindset, even when our whole team fitted around one table in Caerphilly,” said Ruth Kearney, CEO of Nightingale HQ. “Meeting John and Iain reinforced the strength of the relationship between Wales and Ireland and the opportunities that exist for Welsh businesses to grow internationally. That support is invaluable as we continue to scale.”

Looking ahead

We are continuing to expand our presence across Ireland, building on relationships with the wider Irish manufacturing community. As we grow, we look forward to working closely with Welsh Government in Ireland to develop new partnerships, reach more manufacturers, and strengthen the connection between two nations with a shared ambition for innovation and industrial growth.

Further reading

• Nightingale HQ teams up with Welsh manufacturing clusters across North and Mid Wales
• Business Wales features Nightingale HQ as a steel manufacturing success story
• Nightingale HQ becomes first UK start-up to join EIT Manufacturing consortium
• Midland Steel case study
• Less Paper. More Metal. GoSmarter pitches to Women Angels of Wales

Caerphilly, Wales — 9 July 2026 Nightingale HQ, the company behind GoSmarter, has successfully closed its angel investment round, securing the investment needed to accelerate growth across the UK, Ireland and India. The final investment comes from Source Code Control India and is accompanied by the appointment of its founder, Gourav Tandon, to the Nightingale HQ Board of Directors.

This investment adds to £200,000 in non-equity funding from the Welsh Government’s SMART Flexible Innovation Support (SFIS) programme, which is supporting the company’s growth plans over the next 12 months.

Founded in Caerphilly in 2018 by Ruth Kearney and Steph Locke, Nightingale HQ develops GoSmarter, an AI-powered end-to-end compliance and traceability platform for the metals industry. The platform automates data capture, eliminates manual paperwork, and helps manufacturers improve compliance, reduce waste, and free up skilled teams to focus on higher-value work on the shop floor.

The completed funding round provides the company with the runway to expand internationally, building on its existing customer base in the UK and Ireland and its growing presence in India.

Strengthening the Board

Gourav Tandon founded and scaled the India operations of Source Code Control, a specialist consultancy helping organisations manage open source software risk and strengthen software supply chain security. He is widely recognised for his expertise in software supply chain security, digital transformation and cyber security consultancy.

Having built and grown a technology business across the Indian market, Gourav brings valuable commercial experience to Nightingale HQ as the company begins its next phase of growth.

“This comes down to the ability of the Nightingale team to execute,” said Gourav Tandon. “Steph and Ruth are building something genuinely innovative in a massively traditional space. The traction and the opportunity are significant, both globally and especially in India where production capacity is expected to reach 300 million tonnes per annum by 2030. I’m excited to work with them and help them scale.”

A Strategic Partnership

For Nightingale HQ, Gourav’s investment is about more than capital. “We’re delighted with the investment, but it’s our partnership that we really value,” said co-founders Ruth Kearney and Steph Locke. “He brings both vision and execution, and his experience will be invaluable as we expand into the Indian market.”

Building on Early Success in India

GoSmarter is already running trials with metals manufacturers in India, where businesses face many of the same operational challenges as those in the UK and Ireland. Manufacturers continue to lose valuable production time to mill certificate processing and lack of end-to-end traceability that GoSmarter automates using AI.

Gourav’s appointment brings proven experience of scaling technology businesses in India, together with deep expertise in software supply chain security and enterprise technology. His knowledge and network will help Nightingale HQ convert successful trials into long-term commercial growth while strengthening the company’s governance as it expands internationally.

Looking Ahead

Closing the angel investment round marks an important milestone for Nightingale HQ. The company will use the funding to accelerate product development, grow its customer base across the UK and Ireland, and expand its presence in India, where demand for AI-powered manufacturing tools continues to grow. With Gourav Tandon joining the Board, Nightingale HQ is well positioned to support more metals manufacturers in improving productivity, reducing waste and modernising shop floor operations.

About Nightingale HQ

Founded in 2018 and headquartered in Caerphilly, Wales, Nightingale HQ develops GoSmarter, an AI-powered end-to-end compliance and traceability platform for the metals industry. The platform automates data capture, eliminates manual paperwork, and helps manufacturers improve compliance, reduce waste, and free up skilled teams to focus on higher-value work on the shop floor.

Media Contact

Nightingale HQ hello@nightingalehq.ai

A digital twin gives you a live model of the line, so you can spot where flow is getting stuck and test fixes before you touch the plant. For metals manufacturers, that means using live machine data, routing rules, heat numbers and planning records to check line flow on screen first. If you already use GoSmarter, tools like Production Planner, MillCert Reader and Product Lineage help keep that input data clean enough to stop the model turning into another software fairy tale.

What you’ll get here:

• A plain-English view of how digital twins help find moving bottlenecks
• A short take on why spreadsheets miss blocked and starved stations
• A practical starting point for a pilot on one line or one queue
• A simple way to measure throughput, scrap, lead time and £ savings

Here’s how to fix it.

Digital Twins & Industrial Artificial Intelligence Applications for Manufacturing Workflow Efficiency

Where Your Line Actually Gets Stuck

In metals plants, the choke point usually shows up in the flow. Blocked cutting lines. Starved finishing stations. Crane delays. Slow changeovers. Inspection queues. Each one gums up the line in its own way, but the result is the same: throughput drops and the queue gets shoved downstream.

A machine is blocked when it has finished its cycle but the next station or buffer is full. In plain English, it’s waiting for the next step [3]. It is starved when it sits idle because upstream parts have not arrived. That means it’s waiting on supply from earlier in the line [3]. Fix one upstream issue and the bottleneck often pops up somewhere else, usually inspection or material handling [4].

Frequent bottlenecks mean overtime, expediting, and weaker delivery performance [6].

This is where plenty of plants get caught out. They tidy up one station, pat themselves on the back, then wonder why the line still feels slow. The problem is simple: pushing one station harder can dump more pressure on the next one and make overall flow worse [4]. A machine can look busy and healthy on its own while queues are stacking up somewhere else. Line performance matters more than machine uptime. That line-level mismatch is exactly what a digital twin shows before you touch the plant.

Why Spreadsheets Break Down When the Shop Floor Gets Busy

The problem with manual tracking is timing. Spreadsheets are static. They show fixed time periods, not what’s happening now. That becomes a mess when cycle times change with product mix, a maintenance window runs over, or one crane is being pulled across several bays at the same time. A static log shows what happened. It does not show where the next constraint is forming.

The job is not just to record delays. It’s to see how one delay knocks the rest of the workflow out of shape.

How a Digital Twin Shows You the Problem Before You Touch the Plant

A factory digital twin mirrors your machines, material flow, labour, routing rules and queues using live production data. In a metals plant, that means you can see how work moves from furnace to dispatch. The model pulls in heat numbers, mill certificates, cutting plans, scrap records and routing rules across the line.

It shows where queues are building, where stations sit idle and where delays start rippling downstream. And let’s be honest, most plants don’t have one fixed bottleneck sitting there forever. The live constraint shifts with product mix, maintenance and demand. Once you can see that live constraint, you can test fixes on a screen instead of gambling on the shop floor.

That is the whole point. You try changes before anyone touches the plant. Change a shift pattern, add capacity or tweak batch sizes, and the twin shows the knock-on effect on throughput across the whole line.

What Data You Need to Build a Twin That People Will Trust

To make the model useful, start with the records your plant already has. A twin only works when the inputs are clean: process rules, machine signals and traceable material records.

PDF mill certificates, heat numbers and scrap records often live in separate spreadsheets. That mess makes it harder for engineers to trust the twin. GoSmarter helps metals manufacturers turn that scattered data into clean, model-ready inputs by digitising mill certificates, linking heat codes to inventory, and pulling scrap and order records into planning data you can actually use.

Why Legacy Systems Are Not a Barrier to Getting Started

You do not need a full system rip-out to begin. UK plants can start using digital twin methods without replacing the infrastructure they already rely on. A useful twin can sit on top of your current setup, pulling data from older supervisory control and data acquisition and control systems without forcing a full replacement.

The sensible place to start is with the records already sitting in your business: equipment specs, process spreadsheets and CAD exports. Build the model around one constrained area, check it against actual output data, then expand once it matches what the plant is doing.

Fix the Bottleneck on Screen First, Not on the Shop Floor

Once you trust the model, use it to test fixes before you touch the line. Trialling changes on a live production line burns cash fast. A shift pattern tweak that looks fine on paper can clog a furnace queue within hours. A buffer change that seemed safe can leave a downstream station short by the end of the shift. A digital twin cuts that risk. Your operations team can test schedule, staffing, routing and layout changes in a virtual line, then see the flow impact before making changes on the floor.

The idea is simple: test the change virtually, then roll out the version that improves flow.

3 Workflow Problems You Can Test in a Digital Twin This Week

The first is a casting bottleneck. If a caster processes 7,200 tonnes per day while the blast furnace is producing 8,000 tonnes per day, the caster becomes the constraint and the queue starts building straight away [1]. The twin lets you test whether an extra shift, a resequenced batch order or a buffer change clears that queue, without stopping the line just to learn the hard way.

The second is furnace loading patterns that create idle time. If a reheating furnace delivers 520 tonnes per hour but the hot strip mill is rated at 600 tonnes per hour, the mill sits idle for about 12 minutes every hour [1]. Simulating different loading sequences and dwell times helps you find the pattern that keeps the mill fed without overloading the furnace.

The third is internal transport delays between bays. A finishing line processing 40 coils per shift cannot hold that pace if the cooling bed only stages 34. The queue pushes back into the mill and forces holds [1]. Testing a revised transport schedule or a staging layout change in the twin helps you sort that out without knocking throughput on the floor.

These tests show which change lifts throughput without just shoving the bottleneck somewhere else.

What Happens to the Bottleneck After You Fix the First One

This is where plenty of teams get caught out. Fix one constraint, and the next one shows up in inspection, dispatch or material handling. That is not failure. That is exactly what should happen. It has a name: constraint migration [1].

“The plant’s understanding of its own bottleneck is wrong approximately 60% of the time… because the bottleneck moves. It moves when the product mix changes. It moves when maintenance takes equipment offline.” - John Mark, Industry Expert [1]

The twin earns its keep here because it shows you the next constraint before you commit to the first fix. Operations directors can use that view to plan staged improvements in sequence, balancing upstream and downstream capacity instead of patching isolated pain points. Put bluntly: the twin shows the next constraint before you spend on the first fix.

That sets up a live test using plant data.

Run One Live Workflow Test Using GoSmarter Data

Use the next constraint your twin shows you to pick the first pilot. Scheduling is usually the fastest place to start because you can fix flow without buying new kit. In practice, scheduling optimisation can recover 2–4% throughput just by resequencing the work you already run [1].

Start where sequencing choices create the biggest queue. Export your current cutting plans and sequences into a simple single-line model. Map material flow, cycle times, and capacity limits. Nothing fancy. Just enough to show where the line gums up. A single-line pilot usually takes three to six months to give you action you can use [1].

If your team still keys in mill certificates by hand, that’s old-school admin pain you don’t need. Use GoSmarter’s MillCert Reader and Product Lineage to digitise heat codes and tighten traceability. That helps with grade-change checks and can cut scrap by up to 20% [7].

Then run a live test:

• Put a batch of mill certificates through GoSmarter’s MillCert Reader
• Time the manual entry process
• Track how many fields need fixing after entry
• In parallel, use Production Planner data to test a planned 72-hour outage on the bottleneck
• Watch how queues shift and how recovery time changes [1][7]

Measure queue length, correction rate, and recovery time against your current baseline. That gives you a clean before-and-after, not hand-waving.

Conclusion: Start Small, Measure Clearly and Scale What Works

Factory workflow problems stay hidden because the data is scattered, and the bottleneck keeps shifting. One week it’s the saw. Next week it’s packaging. Then it’s the press line because a job overran and nobody spotted it early enough. A digital twin shows you where the constraint is before you start changing the plant.

Once you can see the constraint, the next step is simple: does fixing it pay back, or are you just moving the mess around?

The numbers are hard to ignore. Over 90% of digital twin deployments return more than 10% Return on Investment (ROI), with more than half beating 20% [2]. A three- to six-month pilot on one bottleneck in a single critical line gives you clean before-and-after data on throughput, lead time, scrap and £ savings [1]. That gives you proof, not guesswork.

That makes the first live pilot practical, not theoretical.

For UK metals manufacturers, the path is pretty direct. Pick one bottleneck where your gut says there’s a constraint, but the numbers still don’t prove it. Feed the model clean production data: accurate timestamps, real cycle times and actual mean time to recovery figures. Then measure the result against your current baseline in terms you already care about:

• tonnes per day
• lead time in hours
• scrap as a percentage
• the £ value of the change

If you use GoSmarter’s MillCert Reader and Production Planner, use them to keep certificate data and scheduling inputs clean while you build that baseline.

Once the first win is documented, the twin becomes a shared reference point for planners, engineers and operations leads. It cuts down the usual debates driven by incomplete data and half-finished spreadsheets [4]. Start with one bottleneck. Measure clearly. Then scale what works.

FAQs

Morning Session: Practical AI Training with Steph Locke

The morning session was led by Steph Locke, Co-founder and Head of Product at Nightingale HQ. Steph introduced participants to prompt engineering and practical ways to use AI in day-to-day work. She also shared a simple framework for understanding different levels of AI adoption, from AI acting as an advisor, through assistant and agent roles, to more autonomous forms of automation.

Participants got hands-on with practical exercies to apply AI to their own areas and workflows, they solved business challenges including;

• Improving RFQ turnaround times
• Strengthening Master Order Book management and cashflow visibility
• Enhancing production planning and stores efficiency
• Streamlining documentation and compliance processes
• Accelerating the creation of HR onboarding and training materials
• Eliminating manual searches by enabling instant lookup of component drawings and part numbers
• Reducing administrative workload and improving access to business-critical information
• Increasing efficiency across day-to-day operations through practical AI applications

The focus was on achieving useful outcomes quickly. Teams were encouraged to experiment, refine prompts, and identify opportunities where AI could remove repetitive work and free employees to focus on higher-value activities.

Afternoon Session: Building an AI Strategy with Florence

The afternoon session shifted focus from operational improvement to strategic planning. Led by Ruth Kearney, CEO of Nightingale HQ, Philtronics’ management team used Florence, Nightingale HQ’s new AI Strategy Agent, to identify where AI could create measurable business value and prioritise future initiatives. Rather than producing a lengthy strategy document, the session focused on identifying practical opportunities and creating a roadmap that could support ongoing digitalisation efforts.

This approach helped connect frontline learning from the morning session with leadership decision-making, ensuring that ideas generated during training could be translated into actionable business priorities.

Philtronics CEO Reflects on the Day

Simon Pritchard, CEO of Philtronics, highlighted the practical impact of the workshop:

“Today we ran a practical AI workshop with Steph Locke and Ruth Kearney from Nightingale HQ Ltd for 24 people across Philtronics. The team focused on real challenges — RFQs, documentation, Master Order Book, production planning, build notes, stores efficiency, and HR onboarding. The core question: ‘How can AI help you right now?’ The highlight was watching teams create useful outputs in just 30 minutes. Don’t be afraid to fail and that the biggest gains came when people experimented boldly and learned from imperfect first attempts. Confidence grows quickly once you treat failure as part of the process.”

His feedback reflected one of the key themes of the day: AI delivers the greatest value when it is applied to existing operational challenges rather than treated as a standalone technology initiative.

Continuing a Digitalisation Journey

This workshop was not Philtronics’ first step into digital transformation. Nightingale HQ has worked with the company for several years, supporting process improvement and digitalisation initiatives as the business has grown. The AI workshop represented the next stage in that journey, helping the organisation explore how emerging technologies can complement existing systems and processes while building confidence across the wider team.

Supporting AI Adoption in Manufacturing

The same practical approach has also been used with larger manufacturers, including Tata Steel UK. At Tata Steel, Nightingale HQ delivered AI readiness and capability building programmes across leadership and operational teams, helping employees understand how AI can support business objectives while developing the confidence to use new tools responsibly.

Whether working with a major steel producer or a growing Welsh electronics manufacturer, the approach remains consistent:

• Start with real business problems
• Give people practical experience with AI tools
• Build confidence through hands-on learning
• Connect operational improvements with strategic priorities

Backed by the Welsh Government Flexible Skills Programme AI Pilot

These AI programmes were supported through the Welsh Government Flexible Skills Programme AI Pilot, helping businesses access practical AI training and identify opportunities for productivity improvement.

By combining skills development with real-world application, programmes like this help organisations move beyond pilots and to driving measurable value from AI across their operations. For manufacturers considering AI adoption, the lesson from Philtronics is simple: start with the challenges your teams face every day, give people the tools and training to experiment, and focus on practical outcomes that deliver value quickly.

If you want to run something similar with your own team, get in touch with Nightingale HQ.

This is no longer a proposal to plan around. It is the live rule set your next purchase order runs under. The old safeguards expired on 30 June, and the tougher regime replaced them the same day. Downstream manufacturers who use steel are not happy about it, and their reasons are worth understanding before you sign your next contract.

Here is what actually changed, who carries the cost, and the practical moves to make this week.

What went live at midnight on 1 July

Two separate measures started today. They rhyme, but they are not identical, and you need both if you trade across the UK-EU border.

The old UK and EU steel safeguards both expired on 30 June 2026. Each carried a 25% duty on over-quota imports. A stricter quota-plus-tariff system replaced them from day one of July.

The headline change is the same on both sides. Go above your tariff-free quota and the duty rate is now 50% by value, up from 25%. That single switch can turn a profitable order into a loss if your material lands on the wrong side of a quota threshold.

We covered the build-up to this in our earlier post on the UK doubling steel tariffs and the EU vote to renew its safeguards. This post is the update for the day it all became real.

What the UK measure does to your quotas

The UK measure targets steel products that can be made in the UK. If a British mill can supply it, the quota door is now much narrower.

The new tariff-free import quotas are cut by 51% compared with the old safeguard measure, according to GOV.UK. That is not a trim. It is roughly half your duty-free headroom gone. The out-of-quota tariff then rises to 50% on anything above the line.

Quotas run quarterly and work first come, first served through HMRC. Unused quota rolls into the next quarter within the same year. In plain terms, early buyers get the duty-free tonnes and late buyers pay the 50% penalty. Timing is now a cost lever, not an afterthought.

There is one important escape hatch. Goods under a contract agreed before 14 March 2026, imported between 1 July and 30 September 2026, are fully exempt from the 50% duty. They do not count against the quota either. If you have older contracts still landing this quarter, that transitional relief is worth real money. Find those contracts today.

First come, first served means your quota position depends on when material clears customs, not when you ordered it. Track the clock, not just the price.

What the EU measure changes for cross-border trade

The EU measure follows the same logic with a sharper twist on origin. The tariff-free import quota drops to 18.3 million tonnes a year, a 47% cut against 2024 import levels. Above that, the duty is 50%.

The bigger structural change is the new “melt and pour” rule. A steel product’s country of origin is now where the steel first turned from liquid to solid, as slabs, billets, or ingots. Where final processing happened no longer decides origin. This closes the loophole where a third country lightly finished steel and re-routed it under a friendlier declared origin.

Melt and pour raises the stakes on traceability. Your customs position now depends on documents most teams treat as filing, not data. If your mill certificates do not clearly evidence where the steel was cast, you cannot prove origin quickly when it matters.

For UK exporters there is some relief. Around two-thirds of UK steel exports to the EU stay tariff-free for five years under the UK-EU arrangement. That protects a big slice of cross-Channel trade, but it does not cover everything, and it does not help you on the import side.

Why downstream manufacturers are angry

The reaction to today’s changes is loud, and it comes from the people who use steel rather than make it. The maths behind that anger is simple.

Downstream steel-using industries employ around 300,000 workers across construction, automotive, aerospace, and fabricated metals. Primary steelmaking employs around 30,000. That is roughly ten downstream jobs for every one in the mills. When import costs rise, that larger group absorbs the hit through higher input prices.

The British Chambers of Commerce warned the changes could add millions of pounds to manufacturers’ costs. It also flagged “real financial and logistics problems” for downstream industries. Manufacturers say they face extra costs once quotas run out. Some warn they may have to halt production where specialist steel grades are not available from UK mills.

That last point is the one that hurts most. Protection helps primary producers. It does nothing for a fabricator who needs a grade no domestic mill rolls. Downstream users wanted less intervention and secure supply, not a tighter quota and a 50% cliff edge.

UK versus EU: the two regimes side by side

The two measures share a shape but differ in the detail that decides your exposure.

Read both columns if you trade across the border. A single-country view will miss half your risk.

What to do this week

You do not need perfect forecasts to act. You need to know your exposure and move before the quota clock runs down.

• Find every pre-14 March contract landing before 30 September. Those UK imports dodge the 50% duty and the quota entirely. This is free money if you claim it.
• Map your steel buys by grade, origin, and quota band. You cannot quantify exposure you cannot see. If that view lives in scattered spreadsheets, pull it into one place now.
• Check where your steel was melted and poured. Under the EU rule, origin now hangs on your mill certificates. Weak traceability turns into customs delay at the worst moment.
• Reprice any live quote with post-July delivery. A quote built on last month’s steel cost may already carry a loss. Check every open quotation this week.
• Watch the quota fill rate, not just the calendar. First come, first served means duty-free tonnes disappear early. Order ahead on your highest-volume grades.

The teams that get burned will be the ones still working from last Friday’s stock figures. Volatility punishes bad data faster than it punishes bad luck.

Turn quota chaos into a controllable number

The single most useful move today is to see your steel position clearly: what you hold, what it cost, and where it came from. That is a data problem, and it is the one GoSmarter was built to fix.

GoSmarter, built by Nightingale HQ, sits on top of your existing ERP, spreadsheets, and email. There is no rip-and-replace. GoSmarter’s Metals Manager tracks stock the way a steel business actually works, by length, grade, and heat number, not units on a shelf. That lets you reprice quotes against real cost-per-tonne while input costs move, instead of guessing. When quota bands and the melt-and-pour rule make origin a customs question, the Mill Certificate Reader pulls heat numbers and origin data off your certs automatically. You prove where steel was cast without a paper hunt. Your data stays on UK Azure, and we never train models on it. Steel stockholders can see how the pieces fit together on the steel distributor software hub.

Start with your live quotes and your pre-14 March contracts. Those two lists decide how much of today’s 50% duty you actually pay.

Frequently asked questions

Sources: UK’s steel trade measure from 1 July 2026 (GOV.UK) and British Chambers of Commerce.

The pain is obvious. Bad schedules, thin quotes, missing certs, and people wasting hours digging through PDFs, emails, and mystery fields like ATTRIBUTE1.

I see the fix as pretty simple. You do not need to rip out a 10 to 20-year-old ERP to get useful AI. You need a context layer that pulls in live stock, routings, pricing rules, heat data, and documents before the model answers. That’s where GoSmarter fits for metals manufacturers. Tools like MillCert Reader, Product Lineage, Business Manager, and Production Planner give the AI the facts your shop runs on.

What you get is plain enough:

• Less guessing in scheduling and quoting
• Fewer manual checks on mill certs and traceability
• Lower risk when CBAM and batch records need backing up
• Faster replies to customers without making up lead times or scrap rates
• A sane first step that starts in read-only mode, not a full ERP circus

The article’s point is blunt: a smarter prompt will not fix missing shop-floor facts. If the AI cannot see the machine status, stock position, cert PDFs, or customer rules, it fills the gaps and hopes nobody notices.

Here’s how to fix it.

What the future holds for software development in metal fabrication

Where Simple Prompting Breaks: Schedules, Quotes and Compliance

The cracks show up first in scheduling. Then quoting. Then compliance. Same problem every time: the AI sounds sure of itself, but it doesn’t have the facts in front of it.

Why your production schedule is still a guess

Ask an AI, “What should we run today?” and give it nothing else, and it’ll hand back something that looks sensible. That means very little on a metals shop floor. Sensible-sounding is not the same as right.

Your ERP stores records. It does not always show the live picture. If the model can’t see machine status, coil widths, due dates and changeover times, it starts filling in the blanks. So you get a schedule that is wrong, delivered with total confidence. That is useless for live shop-floor decisions [2].

The gap hits the same places every day:

The same blind spot turns fast quoting into expensive guesswork.

Why fast quotes go wrong when the AI cannot see the rules

A metals quote lives or dies on the details: alloy grade, thickness tolerances, yield assumptions, scrap rates and customer-specific pricing rules. Miss one, and the margin is wrong before the quote even leaves the desk. A prompt-only AI will happily invent the missing inputs with plausible numbers. Looks tidy. Sounds smart. Still wrong for your grade, your customer and your stock at that moment.

A quote built on assumed scrap rates is not a quote - it is a liability with a reference number.

A lot of built-in ERP AI tools make this worse because they can only see what’s on the user’s screen right then. They can’t check inventory, purchase orders and production schedules in one query [6]. So the quote comes back fast, polished and expensively wrong.

Compliance falls over for the same reason. If the model can’t see the certs or carbon data, it can’t check them.

Why scrap, certs and CBAM checks get messy fast

Scrap tracking and compliance look simple until you ask the AI to trace an offcut back to a heat code and match it to a mill certificate PDF. From a vague prompt, with no document access, that’s fantasy.

Get the policy context wrong and you get the carbon figure wrong. Then the filing is wrong too.

What Good Context Actually Looks Like in a Metals Shop

Useful context in a metals shop is not some magic prompt trick. The problem is usually missing shop-floor detail, not weak AI. Pair legacy ERP with the right operational inputs and it stops acting like a dusty filing cabinet. It starts helping you make decent calls.

The 6 pieces of context that change the answer

These are the inputs the model needs before it can make a call.

• Job routings: show the order of operations and where the bottlenecks sit.
• Coil and heat data: confirm whether the material in stock is fit for the job.
• Live inventory status: show what is actually available right now, not what the system said yesterday.
• Pricing rules: apply customer discounts, LME surcharges and freight rules.
• Customer specs: spell out tolerances, packaging and quality rules.
• Production constraints: show capacity, setup time and machine downtime.

Together, these inputs make up the context layer the model needs before it can reason properly.

What a better request looks like on the shop floor

The gap is not prompt length. It is input quality. A longer prompt full of waffle still gives you a bad answer if the model cannot see the facts that matter.

You see the difference fastest in everyday shop-floor jobs:

Context-rich requests give the AI the facts it needs, so the answer matches your shop, your stock and your constraints.

How to Add Context Without Replacing the ERP

Most manufacturing SMBs are still stuck with ERP systems that are 10 to 20 years old. And let’s be honest, ERP migration projects have a nasty habit of chewing through time and cash, then missing the mark anyway. A better move is to add a context layer on top of what you already run. That layer pulls in the right records, documents and rules before the AI tries to answer. The hard part is not replacing the ERP. It’s getting the right context into the answer without tearing the whole place apart.

Use retrieval and document layers to stop the guessing

Retrieval-Augmented Generation, or RAG, pulls the exact record first, then builds the answer from that source. That could be a mill certificate linked to a heat code, or the pricing rule behind a surcharge [4][2]. No guessing. No made-up filler.

This works well with messy data, which is what most old ERPs are full of. Bent fields, odd naming, half-finished records, PDFs nobody wants to touch. That’s normal. You do not need a spotless data warehouse before you start.

Give each team its own AI assistant, not one bot for everyone

One assistant for planners, quality and sales sounds neat on paper. In practice, it’s useless. Each team needs different context.

• Planners need sequence logic and capacity data.
• Quality teams need mill certificates and heat traceability.
• Sales needs live pricing and lead times [7].

Role-based assistants fix this by keeping each user inside the records they actually need. A quality engineer gets mill certificates and traceability data. A sales rep gets pricing rules and lead times. Neither sees the other’s data unless their role allows it. That keeps answers tied to the facts that matter for that job and improves output quality [2]. So yes, the assistant should be scoped by team, not dumped on the whole business as one generic bot.

Where GoSmarter fits when your system is older than your apprentices

GoSmarter sits on top of your current ERP instead of ripping it out. It adds the context machinery that older systems were never built to handle.

• MillCert Reader uses AI OCR to read and digitise PDF mill certificates, cutting manual data entry errors.
• Product Lineage handles heat-code traceability and links inventory records to the right certificates automatically.
• Business Manager covers inventory tracking, order management and scrap tracking in one place, so the AI can see live stock status.
• Production Planner gives you first-draft cutting plans tied straight into inventory and orders.

Rollout matters as much as the software. Start in read-only mode. Let the AI extract and surface data without writing back to the ERP. That gives your team time to trust it and shows up data gaps without putting operations at risk [3][5]. Most shops move from that first phase to AI-assisted recommendations within 8 to 12 weeks [5]. The safest first step is read-only extraction on one job that’s hurting today.

Start Small: Run GoSmarter on One Painful Job This Week

Once the context layer is in place, don’t roll it across everything at once. That’s how good ideas get buried under meetings, setup faff and “we’ll come back to it next month”. Prove it on one live job first.

Pick the single job causing the most pain right now. That might be certificate handling, traceability, scrap tracking or manual rekeying. Go for the smallest live task that shows the gap fastest.

If mill cert handling is the main bottleneck, run MillCert Reader on the next batch of certificates that lands on your desk. If traceability is the bigger headache, use Product Lineage on the next check. Don’t overthink it. Use the job that puts the current process under the harshest light.

The goal this week is simple: prove, on one live job, that context-aware AI beats the current process.

Track what changed:

• time saved
• fewer manual checks
• faster customer reply time
• lower compliance risk
• less rework

Then use that baseline to pick the next job.

FAQs

The mess is familiar. Breakdowns shout louder than prevention, stock is missing when you need it, and half the plant history lives in someone’s head until they leave.

I’d fix it by starting small and keeping it blunt. Use GoSmarter to pull production, stock and job data into one view, then use that view to plan work on the assets that keep hurting output. This is for production managers, maintenance managers and engineers who are tired of chasing faults instead of stopping them.

You’ll get:

• a plain way to spot the Top 5 assets draining labour and uptime
• a tighter PM routine based on repeat failure data, not old paperwork
• a way to stop stockouts, repeat call-outs and patch jobs
• a simple path to use condition checks and AI only where they stop line losses

Here’s how to fix it.

AI in Manufacturing: Predictive Maintenance for ROI & Uptime

Find out where your breakdowns are actually coming from

Turn gut feel into proof. Without clean, structured data, your priorities get set by memory. Then the same faults keep coming back.

Start with the machines that cost you the most when they stop

Go after the worst offenders first. In most plants, a small group of assets causes most of the downtime and repair spend [6][7].

Score each asset against four simple checks:

• Line impact: does it stop the line dead, or just slow things down?
• Safety and environmental risk: does a failure create an immediate hazard?
• Quality impact: does it create scrap or rework?
• Repair time: how long do parts and the fix take?

That gives you a plain three-tier ranking. Tier 1 assets are line-stoppers or safety-critical machines. These need preventive or condition-based maintenance. Tier 2 assets hit capacity or quality but do not stop the line. Tier 3 assets are a nuisance, but they do not hit output directly. Let those run to failure and spend your time where it counts [3][6].

Start with your Top 5. You already know the machines. They’re the ones that keep dragging your team into emergency call-outs. Check the last 6 to 12 months of work orders and find the assets with the most unplanned repairs and repeat failures [6].

Track downtime without creating more admin work

Once you know which assets matter, you need steady data on what is stopping them. The tool matters less than the habit. Use whatever the shift will actually log in. Just make sure every Tier 1 stoppage gets logged.

For each event, capture six fields:

• start time
• stop time
• duration
• machine ID
• cause
• whether production stopped

Keep the cause codes simple: Mechanical, Electrical, Operator Error, Material Shortage. Write them how people on the shop floor speak. “Bearing noise” beats some bloated office label every time, because each shift will log it the same way [9].

If you can’t see the repeat fault, you can’t stop the repeat call-out.

Set a minimum threshold of five minutes for manual logging so you don’t drown in micro-stop clutter. Start with one critical line for 30 days. Find the stoppages. Then roll it out further [7][3]. That gives you enough to build PMs around actual failures, not guesswork.

The difference between guessing and knowing

Without structured tracking, maintenance priorities get decided by memory, habit, and whoever had the worst week. Proper logging shows which machine burns the most time, which fault keeps coming back, and which fix pays back fastest.

62% of small to mid-size facilities cannot accurately identify their top three root causes of downtime [7]. Inadequate preventive maintenance and unresolved chronic failures together account for 62% of all industrial downtime [8]. You cannot fix what you cannot see.

After four weeks on one line, the pattern is usually plain enough to act on. The next section shows how to turn that into PMs and operator checks your team will stick with.

Build PMs and operator checks that people will actually follow

Use the failure patterns you’ve already logged to cut the PM list down to the work that stops repeat breakdowns. That gives you a preventive maintenance schedule based on what failed in the real world, not on guesswork or old habits [6][3].

Write PMs around real failure modes, not low-value tasks

Most PM programmes end up bloated. Tasks pile up. People keep doing them because they’ve always been there, even when they catch nothing. Meanwhile, the checks that stop the same fault happening again get missed.

Don’t add strip-down work unless it stops a known failure. If a critical asset keeps dropping out because of bearing seizure, the PM should cover the lubrication check and the right interval. Not some vague “inspect motor” line that tells nobody what to look for. Set PM frequency inside the P-F interval. If the check happens after that point, you’re not preventing anything. You’re just reacting later [11][4].

Then adjust the interval based on what you find:

• If deterioration shows up at every inspection, shorten the interval.
• If six months go by with no findings, extend it [10][12].

Give operators a one-page check they can complete on shift

The same failure data should shape operator checks. That’s common sense. Operators usually spot the early signs first: odd noises, heat, leaks, frayed belts, warning lights. Autonomous maintenance turns that from “someone noticed something last Tuesday” into a simple routine people can do on shift [11][13].

Keep the checklist short. One page. Use pass/fail limits that leave no room for debate. For example, don’t write “check belt condition”. Write “acceptable belt deflection: 10–15 mm” so every shift checks against the same line. Use °C for temperature readings and mm for clearances [3][10].

Tick each item off on its own. Give people a way to attach a photo if something looks off. And if a check fails, the system should open a corrective work order automatically. Otherwise it sits in someone’s head, on a scrap of paper, or in that black hole called “I told someone about it” [10][3].

Schedule PMs for times when the work can actually get done

A PM plan on paper means nothing if the machine’s never free. Checks only work when the asset is available, the parts are there, and someone has the time to do the job properly. So schedule PMs into planned shutdowns or weekly planning blocks, with production in the loop.

Start with Tier 1 assets and protect those PM slots in the weekly plan [3][10]. Assign each task to a named technician, not a team or department. If everybody owns it, nobody owns it.

If PM compliance drops below 85%, don’t jump straight to blaming discipline. First check the plan, the labour, and the timing. Bad scheduling creates bad compliance far more often than bad intent [10][3]. A short 15-minute weekly review of completed versus overdue tasks is usually enough to spot drift before it turns into a mess [3].

Fix the parts problem and stop repairing the same fault twice

The two biggest drains on your day are missing parts and the same fault coming back again. Once your PMs are sorted, that’s usually where the next batch of pain sits.

Stock the spares that save you the most downtime

Not every part needs to sit on a shelf gathering dust. Start with the parts that create the most maintenance wait time: the top 20% of part numbers, spares linked to Tier 1 assets, and anything with a lead time of several weeks [1][3].

A £12 sensor stuck in a supplier’s warehouse can stop a £4 million production line for two days. That’s the sort of nonsense that wrecks a week. Panic buying makes it worse. Emergency procurement adds £215–£540 per order before you even count the part itself [5]. Do that often enough and the carrying cost of a few extra bearings starts to look tiny. For Tier 1 assets, the rule is blunt: order a spare the same day the original goes into service [1][2].

For high-use items and common failure points like bearings, belts and seals, set min and max stock levels with an automatic reorder trigger [1]. The point is simple: don’t delay a PM because a seal or filter isn’t there [3]. Standardise part specs across assets where you can. Fewer variants means fewer SKUs and fewer ordering mistakes [1].

Stocking the right parts saves hours. Standard job plans save minutes on every repair.

Standardise the jobs your team does most often

Every time a technician has to stop and work out the right torque setting, the correct oil grade, or which tool fits a job they’ve already done ten times, you’re burning time for no good reason. A one-page job plan fixes that fast.

For a motor bearing change or a hydraulic hose replacement, the plan should show:

• the exact parts needed
• the tools required
• the safety isolation steps
• the expected duration
• the critical settings, like torque values, clearances in mm, and oil grade

No guesswork. No half-memory from the last shift. A technician doing the job at 02:00 should finish it to the same standard as the senior engineer who wrote the plan.

Get the technicians who do the work to help write these plans. They know which steps get skipped when the pressure is on. They know which tools are never where they should be. Plans built with their input get used [3]. Plans dropped on them from an office usually end up ignored.

Fix the root cause instead of applying another temporary patch

If the same failure keeps coming back, you didn’t fix it the first time [1]. You just bought yourself a short pause. Patch-and-close feels fast. Root-cause fixes stop the repeat call-outs.

Set a hard trigger: if the same asset fails more than twice within 30 days, root-cause analysis has to happen before the work order closes [6]. Use 5 Whys for mechanical failures. Use a Fishbone when the cause might sit across maintenance practice, operator setup, design or training [6][1]. Log the finding with failure codes that split out mode, cause and effect [6].

Then do the bit many teams skip. Close the loop.

• If the root cause is a missing PM, add the check
• If it’s a training gap, update the procedure
• If it’s a design issue, make the modification
• If it’s a repeat failure, require supervisor or reliability sign-off before closure [6]

That one control stops band-aid repairs being passed off as fixed. It also feeds the result back into PM updates and operator checks, so the loop stays tight: log, fix, prevent, fail less often.

That cleaner failure data is what makes condition checks and AI alerts worth using. Once spares and repeat failures are under control, target condition monitoring where it prevents the most stoppages.

Where condition monitoring and AI actually earn their place

Use condition monitoring where an early warning gives you time to act. That’s it. The earlier sections covered PMs, operator checks, spares and root-cause fixes. Monitoring and AI sit on top of that groundwork. They do not replace it.

Use condition checks to catch problems before they stop the line

Rotating assets like motors, gearboxes, hydraulic pumps, fans and compressors usually fail in ways you can measure. That makes them a good fit for condition signals. If vibration starts climbing or pressure starts drifting, you get a window to step in before the line stops. But that only works if someone already owns the alert, knows what to do, and has a set response time.

Condition monitoring falls apart when alerts go nowhere. No owner. No action. No deadline. Just another flashing warning on a screen no one trusts.

Before you fit a sensor, pin down:

• the signal
• the owner
• the action
• the response window

That leaves AI for the assets where a simple threshold won’t cut it.

Apply AI to the assets that can really hurt the plant

Predictive maintenance can cut unplanned downtime. It works best on high-consequence assets with a known failure pattern. If one signal tells you enough, stick with rule-based monitoring. If failure shows up across several signals at once, AI starts to earn its keep.

Start small. Pick one or two critical assets. Run shadow mode for six weeks so you can tune thresholds before any live alerts go out. [2][14]

Once the thresholds are tuned, link alerts to live production and stock data.

Use GoSmarter to connect production data to maintenance decisions

Most maintenance teams don’t have a sensor problem. They have a data trust problem. Work orders, fault codes and stock records don’t line up. One system says the part is in stores. Another says it was used last month. The spreadsheet says something else again.

GoSmarter’s comprehensive inventory and order management capabilities along with our APIs keep your data lined up with your ERP. The production planning for long products shows where planned work fits around live production. Clean production data and clean maintenance data are part of the same mess. Connect them, and your calls get sharper

Use that joined-up view to pick the line for a 30-day pilot. That gives you a clean starting point for the 30-day pilot in the next section.

Your next step: run a 30-day pilot on one line and feed the data into GoSmarter

Now do this on one line only. Not the whole plant. Pick the line that causes the most grief. Start with the one sitting at the top of your downtime log or the one with the most breakdowns. Then sanity-check it with six to twelve months of work order history before you commit.

Track these four metrics every week:

• PM compliance above 90%
• MTBF on Tier 1 assets
• Planned work at 70–80%
• Any PM delayed by a stockout[3][6]

Each Friday, review:

• Overdue PMs
• Stoppages over 60 minutes
• Any asset that failed twice in 30 days[6][4]

Use that weekly review to tighten the operator checklist. If a task adds no value, cut it. Keep the checklist to one page. Use clear pass/fail limits, such as “belt deflection 10–15 mm”[3]. And don’t let this turn into another office-made form nobody trusts. Get the technicians who actually do the work to help write it.

Once the checklist is live, connect it to stock and production data so jobs only go ahead when parts and time are there in the first place. That’s where GoSmarter comes in. Use our intelligent inventory and order management capabilities along with our APIs keep your data lined up with your ERP. You can pull production, scrap and inventory data into one view. Then you can check stock and scheduling before you release maintenance work, instead of finding out too late that the job was dead on arrival.

FAQs

Both documents prove one thing: this batch of metal was tested, and the results meet a recognised standard. Get that wrong and you carry a compliance risk into every job. Below, we sort out what each document is, where it comes from, and why the difference matters less than most people think.

Why these documents matter at all

Every week, your goods-in bay takes deliveries from mills, stockholders, and suppliers. Before any of that metal touches a production run, you need proof it meets the spec on the order. That proof is the certificate.

An MTR or MTC does five jobs at once:

• Confirms the material matches the grade and standard you bought
• Gives you full traceability back to a specific batch of steel
• Provides evidence when an auditor or customer asks
• Cuts the risk of a material failure on a finished job
• Keeps you compliant with customer and regulatory requirements

Skip the certificate and you are taking the supplier’s word for it. That is fine until something cracks, fails, or gets queried. Then you need the paper, and you need it fast.

What a Material Test Report (MTR) actually is

A Material Test Report records the chemical and mechanical properties of a batch of material. It confirms the metal supplied meets a particular grade, specification, or standard. Think of it as the metal’s exam results.

A typical MTR lists:

• Material grade and specification
• Heat number or batch number
• Chemical composition
• Mechanical properties
• Testing methods used
• Heat treatment details
• Manufacturing information
• Inspection and approval sign-off

The report ties the physical bar, plate, or coil on your shop floor back to the production records at the mill. That link is the whole point.

What a Mill Test Certificate (MTC) is, and how it differs

A Mill Test Certificate carries much of the same information as an MTR. The difference sits in who issued it. An MTC comes from the original manufacturer or steel mill, and it includes the actual test results from that production batch.

In practice, most teams treat MTR, MTC, Mill Certificate, Mill Sheet, and Mill Test Report as the same thing. The distinction that genuinely matters is the source and the level of traceability behind the paper. A certificate issued by the mill that made the steel gives you stronger traceability than one re-typed by a stockholder three links down the chain.

If two people in your business call the same PDF by two different names, the document is not the problem. The filing system is.

For formal definitions of MTC, heat number, and the EN 10204 standard, see the metals manufacturing glossary.

The heat number is the part that matters

The heat number is the single most important field on any certificate. It is the metal’s unique identifier. It lets you trace a finished product back to the exact batch of steel it was made from.

Lose the heat number and full traceability falls apart. You cannot prove which batch fed which job. During an audit or a customer inspection, that gap turns into a problem you have to explain on the spot. A certificate without a valid heat number is barely a certificate at all.

What to check on every certificate

When a delivery lands, run through the same checks every time. They confirm the metal you received matches the purchase order and the project spec.

Material identification: grade, dimensions, quantity, and the standards it claims to meet.

Traceability: heat number, batch number, and manufacturer details.

Chemical properties: carbon, manganese, silicon, sulphur, phosphorus, and any other alloying elements.

Mechanical properties: yield strength, tensile strength, elongation, impact test results, and hardness where it applies.

Certification details: inspection type, authorised signatures, the issue date, and the standard references.

Miss one of these and you might accept material that never matched the order in the first place.

Where manual certificate management falls down

Managing certificates is rarely the clean process it looks like on paper. Most teams still run it by hand:

The manual route leaks time everywhere. Lost certificates. Missing heat numbers. Duplicate records. Slow audit prep. Hours of admin that nobody enjoys. When you grow from dozens of certificates a week to thousands a month, that workload stops being an annoyance and becomes an operational drag.

How Artificial Intelligence (AI) handles certificates at scale

Forward-thinking manufacturers now hand the grunt work to Artificial Intelligence (AI). A trained model reads a certificate the way your best engineer would, only faster and without getting bored at 4pm.

GoSmarter’s MillCert Reader does the heavy lifting:

• Extracts certificate data automatically
• Identifies heat numbers and material grades
• Validates required fields and flags anything missing
• Builds a searchable digital library
• Creates an instant audit trail
• Links each certificate to inventory and production records

That cuts the admin and lifts accuracy at the same time. Midland Steel, a rebar manufacturer running operations across Ireland, the UK, and Norway, saved roughly ten hours a month doing exactly this. The full story sits in our MillCert Reader case study. For a deeper walk-through of the workflow, see our mill certificate automation guide.

Less paper, more metal

MTRs and MTCs are not just paperwork. They are the backbone of quality assurance, traceability, and compliance across the metals supply chain. The hard part was never understanding what the documents are. The hard part is managing them at scale without burning a day a week on it.

If your team still files certificates by hand, start with one batch. Run your next set of mill certs through MillCert Reader and see how much of the admin disappears. Time spent searching for a certificate is time not spent making metal.

FAQs

AI Optical Character Recognition (OCR) for metals manufacturers is where the real time savings start. GoSmarter reads messy mill certs, drawings and delivery paperwork, pulls out the specs, heat numbers and grades, then sends the data into the systems you already use. No new Enterprise Resource Planning (ERP) system. No giant IT project. Just less typing and fewer bad numbers creeping into stock, quotes and traceability records.

What you get:

• Less admin sludge in quoting, planning and quality
• Fewer typing mistakes on grades, lengths, tolerances and heat numbers
• Faster cert lookup, with records found in under 30 seconds
• Shorter planning time, with one trial dropping from 2 hours to 15 minutes
• Lower scrap, with a reported cut of 2.5 percentage points
• A clean trial path, starting with MillCert Reader on its own

Here’s how to fix it.

Where manual paperwork hits hardest

Manual retyping does the most damage in quoting, planning and quality. Same workflow. Same paperwork. Different ways to lose time and money.

Quotes slow down when every grade and length must be typed by hand

A sales engineer gets a blurry drawing PDF or a faxed Request for Quotation (RFQ). The grade, tolerance band and cut lengths are on the page. But they don’t jump into your ERP or quoting system by magic. Someone still has to type every field by hand. Manual quote capture takes 10 to 20 minutes, and pulling data from a technical drawing can take up to two hours per drawing [1].

This is where the mess starts.

• Get the grade wrong and you allocate the wrong stock.
• Get the length wrong and the cut plan falls apart.
• Misread the tolerance and you quote work you can’t hold in production.

By the time anyone spots it, the quote has gone out, the customer has said yes, and your margin has already taken the hit.

Those are the same fields AI OCR can pull on its own.

Planning breaks down when work orders start with bad data

Planners spend 20% to 40% of their working time retyping data from PDFs into spreadsheets or ERP systems [1]. That’s dead time. You should be building cut lists, making nesting calls and keeping the schedule from drifting off course. Instead, you’re keying in numbers from a document somebody should never have printed in the first place.

The scrap issue is where this starts burning cash. Industry best practice for scrap rates in long product manufacturing is 2.5% or lower, but manual planning often lands between 3% and 8% [7]. On a busy week of rebar or structural sections, that gap is not small. It’s material cost walking out the door.

In a December 2024 UK trial, Midland Steel processed 734 tonnes across 193 jobs. Planning time dropped from roughly two hours to 15 minutes, and scrap fell by 2.5 percentage points, cutting their previous baseline in half [7][1]. The operations manager noted:

“A morning of planning became a five-minute review.”

• Operations Manager, Midland Steel [7]

Clean input data is what turns scrappy paperwork into a schedule you can actually use.

Quality teams should not spend afternoons typing heat numbers

Manually processing a single mill test certificate takes about 12 minutes [1]. On a busy goods-in shift, that eats a big chunk of a quality engineer’s day. And for what? Typing. Not checking. Not verifying. Just moving data from one place to another like it’s still 1998.

Under EN 10204, ISO 9001 and AS9100, traceability is a compliance requirement. Heat numbers, chemical compositions and mechanical properties must be recorded correctly and ready to pull when needed. If a customer asks for proof that material meets spec, you need that cert fast. An AI-indexed system can retrieve it in under 30 seconds [1].

A single transposed digit in a heat number breaks the traceability chain. That’s not skilled quality work. That’s admin drudgery pretending to be quality control.

These are the exact documents AI can clean up first.

How AI reads messy documents and turns them into usable data

Old-school OCR just reads characters. Fine, if your paperwork is neat, flat and from the same supplier every time. On a shop floor, it rarely is. AI document capture goes further. It works out what each field is, then sends it to the right record.

So when a faxed mill cert turns up skewed, scruffy and laid out nothing like the last one, the system uses layout recognition to find the fields without needing a supplier-specific template. It also maps terms like “Heat No.”, “Cast No.” and “Schmelznummer” to the same field in your records [2][6]. That matters because the same bit of paperwork often feeds quoting, planning and quality. If someone has to retype it three times, you’re paying for the same admin three times.

The fields AI pulls automatically from mill certs, drawings and delivery notes

The output is structured data, not just a tidier image. From a single mill cert, AI can pull:

• heat and batch number
• material grade
• chemical composition
• mechanical properties, including yield strength, tensile strength and elongation
• product dimensions and tolerances [2][4]

That data goes straight into your ERP, scheduling or quality records without anyone touching a keyboard.

At Midland Steel, MillCert Reader pulled chemistry and mechanical properties from incoming certs, saving 10 hours of manual admin a month and renaming documents by heat code [8].

“What used to take hours every week is done in seconds.” - Production Manager, Midland Steel [8]

Getting the fields out is only half the job. The other half is making sure bad data doesn’t slip into ERP and cause a mess later.

How AI checks data before bad data hits production

Extraction is only half the job. AI gives every field a confidence score. If it’s not sure, it sends that field for human review before anything updates in ERP [2]. The high-confidence fields go straight through. GoSmarter flags the doubtful ones for review.

Better systems also run metallurgical plausibility checks. They check whether the total of alloying elements is physically reasonable, or work out the Carbon Equivalent Value (CEV) to see if the data stacks up [6]. That means GoSmarter catches bad heat numbers, wrong grades or invalid chemistry at review. Nothing bad reaches quoting, scheduling or traceability.

Manual typing versus AI capture: a direct comparison

AI extraction hits field-level accuracy of 97% to 99.79% on technical metals documents [2][6]. Manual data entry has no such benchmark. It also gives you no built-in warning when someone keys in the wrong value after a long shift.

That production manager saved 10 hours of manual admin per month [8]. That’s roughly three working weeks a year [5][4]. Not magic. Just less typing, fewer mistakes and clean data ready for the next step.

How GoSmarter removes the PDF problem without replacing your ERP

This is where the clean data from the last section stops being a nice idea and starts doing actual work. GoSmarter’s AI production assistant sits on top of your existing ERP, quoting and quality systems, so the data lands where your team already works. No more printing, typing, checking, then typing it all again because some poor soul missed a heat number the first time.

Start with MillCert Reader and stop retyping cert data

GoSmarter MillCert Reader pulls in PDFs and scans straight from email inboxes, shared folders or scanners. From a single mill cert, it extracts heat numbers, grade, chemistry and mechanical properties [4][5]. That data then links straight to your stock traceability records, so every item in stock stays tied to its heat code. When a customer or auditor asks for a cert, retrieval drops from 20–30 minutes of folder-hunting to under 30 seconds [1][4].

“GoSmarter saves us hours every month - it pulls the key data out of mill certificates automatically and renames the files straight away. That whole process used to be painfully manual.” - QC Manager, UK Steel Stockholder [4]

Once the data is in, you get control as well as speed. Low-confidence reads go to human review. Every correction is logged for audit. Unapproved certs stay in draft until reviewed, which gives you an immutable record for EN 10204 3.1/3.2 compliance [4][3].

Clean input data leads to better cutting plans, less scrap and faster job release

Clean cert data doesn’t just tidy up the back office. It feeds straight into Business Manager, Production Planner and the Rebar & Scrap Optimiser. If the input data is right, your first-draft cutting plans are right. Offcut tracking makes sense. Job release doesn’t grind to a halt because someone’s chasing a missing heat number through three folders and an inbox from 2022.

In Midland Steel’s trial, clean cert data fed cutting optimisation, cutting scrap by 2.5% and reducing planning time from two hours to 15 minutes [1][7]. That takes planning from a chunky morning task to something you can sort before the tea goes cold.

“Turned a morning of planning into a five-minute review. We cut scrap rates in half during trials.” - Operations Manager, Midland Steel [7]

What the numbers look like on a UK shop floor

On a UK shop floor, the payoff shows up in time, scrap and faster release. It also shows up in the problems you don’t have. A single compliance incident or recall avoided can save between £5,000 and £50,000 [7].

Run GoSmarter MillCert Reader on your next batch of certs and measure the hours you get back

If you want to see the savings on your own shop floor, keep it simple. Start with one cert flow. Pick incoming mill certs from one supplier group, set three baseline figures, and run GoSmarter MillCert Reader on that flow for 14 days [4].

Track:

• cert retrieval time
• data-entry errors caught
• morning planning time

Run the trial without touching your current systems. Upload your existing PDFs, index the backlog, then process live incoming certs in parallel with your current spreadsheet for the full 14 days [9]. No big IT circus. No ripping out tools your team still needs. At the end of the fortnight, compare the two.

A team processing 200+ certificates a month can cover its annual subscription cost through labour savings alone. The estimate is £3,600 per year, based on a £30-per-hour labour rate [4]. The trial gives you full access for 14 days at no cost [4].

Start with MillCert Reader on its own. No ERP connection required. That means you can prove the Return on Investment (ROI) against the certs your team is retyping today, then measure the hours you get back from manual entry. If the numbers stack up, expand from there [4].

FAQs

“Traceability shows you where material went. Audit trails prove what happened to the data.”

Why Audit Trails Matter in Metals

Metals manufacturing lives on documentation

The metals industry lives on documentation. A missing heat number or an unexplained MTR edit isn’t just messy admin. It’s a compliance failure, a potential recall trigger, and a very bad conversation with a customer. They have their own regulatory requirements to meet.

Here’s what a proper audit trail actually gives you:

• Proven data integrity — GoSmarter logs every change (who, what, when, why) so records can’t be quietly altered. You get a complete, auditable history of your data, not just your materials.
• Real traceability — not just linking data, but proving the complete history of MTRs, batches, and orders
• One-click audit prep — pull a full change history instead of chasing spreadsheets, emails, and filing cabinets
• Faster incident response — isolate the problem quickly instead of widening a recall because you can’t tell what changed
• Natural accountability — when every action is attributable to a person, errors drop without anyone needing to say a word
• Process insight — repeated corrections signal poor processes or bad data upstream; you can only see the pattern if it’s logged

What Happens Without One

Let’s be honest about the alternative.

Without audit trails, MTR values get edited with no record of what the original said. Inventory adjustments can’t be explained. Order spec changes go unnoticed until a customer raises a non-conformance. Auditors ask questions you can’t answer. And when something does go wrong, you’re stuck with a wider recall. You can’t narrow down what was affected.

It’s not a hypothetical. It’s the default state of most metals operations. They run on spreadsheets and legacy software built before data accountability was even a purchasing requirement.

The cost isn’t abstract either. A single traceability dispute — a customer non-conformance report (NCR), an internal recall, or a failed audit — typically runs to £5,000 in staff time for a mid-size operation. Quarantine product, and you’re into six figures. An audit trail that takes 90 seconds to pull doesn’t just satisfy the auditor. It limits the blast radius.

The Standards That Require Them

This isn’t optional for most metals manufacturers. Here’s the regulatory landscape:

The FDA and EU GMP standards apply specifically if you supply into pharmaceutical, medical device, or similarly regulated supply chains. For everyone else, ISO 9001, IATF 16949, and AS9100 are the ones that matter.

The common thread across all of them: audit trails must be automatic, time-stamped, secure, and tamper-resistant. “Dave thinks he remembers” doesn’t appear in any of those standards.

What “Good” Actually Looks Like

A compliant, defensible audit trail covers six things:

1. Before and after values

Not just that something changed, but what it changed from and to. “Yield strength modified” is not an audit trail. “Yield strength changed from 355 MPa to 350 MPa” is.

2. User identity — no shared logins

Every action must trace back to an individual. Shared logins make the entire log worthless. GoSmarter enforces individual credentials; every change is attributed to a specific person.

3. Server-side timestamps

Precise, generated by the server, and not editable by the user. A timestamp that can be changed is not evidence of when something happened.

4. Approval / flagging reasong

Optional in some systems, but invaluable during an audit. GoSmarter supports adding a note against approvals or rejections of mill certificates.

5. Immutability — the record cannot be edited

An immutable, tamper-resistant log is what separates a defensible audit trail from a spreadsheet with a “last modified” timestamp. GoSmarter audit log entries are written as append-only records; they cannot be modified or deleted at the database level, including by GoSmarter staff.

6. Full coverage across all data areas

MTRs, cert heats, inventory, orders, order lines, scrap log, and reference data. Not just the obvious bits. Partial trails are fine until the moment they’re not.

GoSmarter vs. Alternatives: Audit Trail Capability

If you’re evaluating tools, here’s what the audit trail landscape actually looks like:

The critical distinction is “automatic” vs. “configurable.” A generic ERP can produce an audit trail if it’s configured correctly and if your team uses it correctly. GoSmarter produces one automatically, from day one, regardless of whether anyone thought to set it up.

GoSmarter Now Does This Across the Board

What GoSmarter logs

GoSmarter’s audit trail isn’t limited to three data types. It spans the full operation. Here’s the current coverage:

Every entry captures the original value, the new value, the user, and a server-side timestamp. That timestamp cannot be edited after the fact.

Deleted items are tracked and recoverable across all the above. For orders, a bulk restore is available and cascades to all line items, so if an order is accidentally removed, everything under it comes back in one action.

Audit trail coverage is available across all GoSmarter plans, including MillCert Reader, which digitises and links your incoming MTRs automatically before the audit trail begins.

How fast can you retrieve an audit history?

Under two minutes. Search by record, user, date, or change type. No chasing spreadsheets, emails, or filing cabinets. If an auditor asks “who changed this MTR value last Tuesday?”, you have the answer in seconds — a complete, auditable history of every change, fully exportable for a customer or regulator.

This is the kind of feature that sits quietly in the background until the day you desperately need it. We built it so that day goes well for you.

The full details are in the release notes. Short version: if something changes in GoSmarter, you’ll know exactly what, who, and when.

See how GoSmarter’s compliance solution maps to each of these requirements. Or explore the full picture of integrated cert traceability and audit trails working together.

Frequently Asked Questions

Further Reading

• GoSmarter Metals Manager — real-time inventory management with a permanent audit trail built in
• Audit Panic is Optional: How to Stop Freaking Out Over Lost Certs
• AI and Mill Test Report Traceability
• MillCert Reader: Automate Mill Certificate Processing
• Case Study: MillCert Reader Saves 10 Hours a Month

You feel it on the shop floor. A bar is sitting in the yard, ready to cut, but nobody can confirm the cert is approved. So it waits. A job slips. The material is fine. The traceability is not. That gap costs you throughput, and it never shows up in a single, neat line on a report.

Most factories treat traceability as a quality job. The Quality team owns the certs, files the Material Test Reports (MTRs, also called Material Test Certificates or MTCs), and dusts them off at audit time. That framing is the problem. Traceability is a production data flow. When it works, your day runs smoother. When it breaks, you lose hours you will never get back.

Here is what end-to-end traceability gives you as an Operations Manager:

• Faster material release, because cert status is visible at the point of use
• Fewer wrong-material incidents, because heat numbers stay linked to jobs
• No audit drills, because the record is already built
• Faster recall isolation, because you can trace one heat in minutes
• Protected margins, because none of the above eats your team’s time

Let’s walk the chain, step by step.

The Chain Is Only as Strong as Its Weakest Handoff

Traceability is not a single record. It is a thread that runs from the mill to your customer. The thread passes through goods-in, stock, job reservation, cutting, despatch, and the customer cert pack. Each handoff is a place where the thread can snap.

Material arrives with a perfect cert. By the time it reaches the saw, nobody knows which heat it belongs to. The chain broke at stocking, and everything downstream is now guesswork.

This is the core point. Perfect traceability at goods-in is worthless if it dies at the next step. A cert filed in a folder while the steel goes into a rack, unlabelled, gives you nothing. Not when a job needs it. You have proof the material is good. You cannot prove which material is which.

Heat number traceability only works when the heat number stays attached at every stage. That is what “end to end” actually means. Not a cert in a drawer. A live link from steel to certificate to job to despatch note.

Goods-In: Where the Thread Starts (or Snaps)

Goods-in is the first handoff, and it is where most chains start with a knot. A delivery arrives. The cert is a PDF, a scan, or a creased sheet. Sometimes from a supplier who has never met a scanner. Someone re-keys the heat number, the grade, and the chemical results into a spreadsheet. Or they do not, and the cert goes in a pile.

Every manual keystroke here is a chance to get it wrong. A transposed heat number means the wrong material is linked to the wrong cert from minute one. That error then travels the whole chain, undetected, until an auditor or a customer finds it for you.

The fix is to verify the cert against the EN 10204 type at goods-in and link it to the physical material straight away. Capture the data once, accurately, and attach it to the stock. The thread is now tied to the steel, not to a folder.

Stock and Reservation: Keep the Heat Number on the Steel

Once material is booked in, it sits in stock until a job needs it. This is where traceability quietly dies in most yards. The certificate lives in your Enterprise Resource Planning (ERP) system or a shared drive. The steel lives in a rack. Nothing connects the two except someone’s memory.

Real traceability links stock to cert by heat number, and keeps that link when material is reserved against a job. When a job is allocated 2 tonnes of a specific grade, the system reserves stock with a known heat and a known, approved cert. The picker does not guess. The cutter does not check three folders. The right material is the easy choice, not the lucky one.

Cutting and Production: One Wrong Heat Costs a Job

The saw does not care about your paperwork. It cuts what it is given. If the wrong heat reaches the cutting list, you do not find out until the part fails inspection, or worse, until it is welded into a customer’s structure.

A wrong-material incident is not a quality statistic to an Operations Manager. It is a scrapped batch, a remade job, an angry customer, and a hole in the schedule. The cost is throughput and rework, not a line in a compliance log.

End-to-end traceability stops this at the cut. The heat number stays attached through production, so every offcut and finished part still carries its origin. If a question comes up later, you trace the part back to the heat, the cert, and the delivery. No detective work.

Despatch and the Customer Cert Pack: Close the Loop

The last handoff is despatch. The order ships, and the customer wants the cert pack. If your chain held all the way through, the pack assembles from records you already have built. If it broke anywhere, someone now reconstructs the trail by hand, before the lorry leaves.

A clean despatch links each item on the order to its heat number and approved cert, then produces the On-Time In Full (OTIF) shipment with its documentation in one go. The customer gets a branded cert pack. You get the despatch out the door. Nobody spends the afternoon collating PDFs.

This is also where recall isolation lives. If a heat is later found out of spec, you trace every order it touched in minutes. Not days. You isolate the affected jobs and contain the problem. A broken chain turns the same recall into a frantic sweep through folders and memory.

Audit Day: Built, Not Crammed

An audit is not a project when traceability runs end to end. The auditor asks for the cert behind a heat number, and you pull it up. They ask who approved it and when, and the log answers. The audit trail built itself while you worked.

Compare that to the usual scramble. Someone spends two hours chasing one cert, then another two on the next request. At £30 an hour, a single audit fire drill burns a day of skilled time on filing. Not steel. Run that against every customer audit and quality review in a year and the cost is real money.

The best audit is the one you do not notice. The record was always there, because every step wrote to it as it happened.

Traceability Is an Operations Metric

Stop treating traceability as a QA cost centre. It belongs on your operations dashboard, next to throughput and on-time delivery. Faster material release is a throughput gain. Fewer wrong-material incidents is a scrap and rework gain. No audit panic is a recovered-time gain. These are your numbers, not the Quality team’s.

The companies that get this build the chain once and run it forever. We saw exactly that pattern in our work with Midland Steel. Joining up processes across the business removed the manual handoffs that slow rebar supply. The principle holds for any stockholder or fabricator. For the full workflow view, our guide to end-to-end mill cert traceability walks every stage from upload to customer pack.

Frequently Asked Questions

Build the Chain Automatically as You Work

The fastest way to make traceability an operations advantage is to stop building it by hand. GoSmarter MillCert Reader builds the EN 10204 audit trail automatically as you work. The AI extracts the cert, your team reviews and approves it, and the system links it to stock by heat number. From there it reserves against jobs and travels through to the despatched order with a branded cert pack.

That is the exact end-to-end chain this post argues for, working without an IT project, a consultant, or a migration. Start a 14-day free trial, no credit card needed, and run it on your next delivery of certs. See the heat number travel from goods-in to despatch on a real job.

For the bigger picture, the integrated cert traceability hub covers how the whole chain joins up, and the mill cert automation hub explains how the extraction works. Then put it on your next batch and watch material release speed up.

This is our second major green-tech recognition in a row. Last year, Nightingale HQ won Best GreenTech at the Wales Technology Awards 2025. Two awards, two judging panels, one consistent message. Practical AI cuts waste in the metals sector.

GoSmarter is built by Nightingale HQ. The product is built for one job. It removes the manual drudgery that slows stainless steel firms down and quietly burns through material, energy, and time.

What the award recognises

The award recognises measurable impact, not promises. Sustainability in stainless steel is not a poster on the wall. It is fewer offcuts, fewer reworks, and fewer tonnes of carbon for every order you ship.

GoSmarter delivers that in four concrete ways:

• Less material waste and scrap. Smarter cutting plans squeeze more parts from every bar and plate, so fewer offcuts hit the skip.
• Lower carbon emissions. Every kilo of stainless you do not waste is carbon you do not pay for, in production and in transport.
• Faster production planning. You plan a batch in minutes, not a morning, so machines sit idle less often.
• Less manual paperwork. The Mill Certificate Reader reads clumsy PDF certs in seconds, so engineers stop retyping and start building.

These are the same outcomes we target in live projects. Our work with Midland Steel shows what happens when planning gets faster and waste gets cut at the same time.

Why this matters for stainless steel firms

Stainless steel carries a real cost. It is energy-hungry to make, expensive to buy, and unforgiving when an order goes wrong. Waste a bar and you waste the carbon baked into it.

That is where GoSmarter’s operations tools earn their keep. They sit on top of your existing process and take the boring, error-prone tasks off your team. You keep your machines and your people. You lose the paperwork and the scrap.

“Sustainability and profit are not enemies,” said Ruth Kearney, CEO of Nightingale HQ. “Cut the waste, cut the carbon, and protect your margins at the same time. The BSSA recognising that work means a great deal to our whole team.”

The point is simple. Greener operations and healthier margins come from the same fix. Stop wasting material, stop wasting time, and the carbon savings follow.

What comes next

We are not slowing down. The BSSA award sharpens our focus on the firms that make stainless steel every day, often with tight teams and tighter deadlines.

If you cut, plan, or ship stainless steel, book a demo and we will show you where the waste hides in your process. Bring a stack of mill certs and a recent cutting job. We will work through real numbers, not slideware.

Our thanks go to the British Stainless Steel Association and the wider community.

Further reading

• Nightingale HQ wins Best GreenTech at Wales Tech Awards 2025
• Mill Certificate Reader
• Midland Steel case study
• BSSA Annual Conference & Dinner 2026 brings the stainless steel sector together in Birmingham

The answers sit inside a single document: the Material Test Report (MTR), also known as a Mill Test Certificate (MTC).

Many organisations treat MTRs as compliance paperwork that tags along with material deliveries. They are much more than that. The MTR proves material quality, and it anchors traceability and compliance across the whole supply chain. Without it, manufacturers, fabricators, distributors, and end users have no reliable way to verify the materials they buy, process, or install.

“Compliance starts with documentation. Operational excellence starts with knowing exactly where your material came from and where it is going.”

What Is an MTR?

A Material Test Report is a quality assurance document issued by the producing mill or manufacturer. It certifies information about a specific batch or heat of material, including:

• Heat number
• Material grade
• Product dimensions
• Chemical composition
• Mechanical properties
• Testing results
• Applicable standards
• Manufacturer certification

The document confirms that the material supplied meets the specification the customer requested and complies with relevant industry standards. In regulated industries, the MTR matters as much as the material itself.

Proof of Material Quality

The primary job of an MTR is to verify quality.

When a buyer orders S355 structural steel, ASTM International (ASTM) A36 plate, stainless steel 316, or any other specified material, they need assurance that the product delivered is exactly what they asked for. The MTR provides that assurance.

Rather than trusting labels, supplier claims, or a visual inspection, buyers review certified test results that confirm compliance with the required standard. Quality teams can then verify:

• Material grade
• Chemical composition
• Mechanical performance
• Manufacturing specifications
• Testing requirements

Skip this verification step and you run a real risk: incorrect, substandard, or non-compliant material slips into production.

The Foundation of Traceability

Quality verification matters, but the real power of an MTR lies in traceability.

Traceability tracks material across its entire lifecycle, from the producing mill to the finished product delivered to the customer. At the heart of that process sits the heat number.

A heat number uniquely identifies a batch of material produced during a specific manufacturing run. Think of it as the material’s fingerprint. It links:

• The physical material
• The manufacturing process
• Laboratory testing results
• Supplier records
• Inventory systems
• Customer deliveries

That connection creates a complete chain of custody. If a problem surfaces months or years later, you can quickly answer:

• Where the material originated
• Which customers received it
• Whether other batches are affected
• What corrective actions you need to take

Without an MTR, none of that visibility exists.

Why Traceability Matters

In industries where safety and performance are critical, traceability is not optional. A single material failure carries serious consequences.

Whether it is a structural component, pressure vessel, pipeline, rail system, offshore structure, or industrial machine, you need confidence that every material used meets the required standard. Strong traceability systems help you:

• Reduce risk
• Improve quality assurance
• Manage recalls effectively
• Respond to customer queries quickly
• Support audits and inspections
• Prove compliance to auditors

Tracing material back to its source protects your organisation and your customers.

Compliance Starts with Documentation

Modern manufacturing operates inside an increasingly complex regulatory environment. Standards such as EN 10204 (the European standard for inspection documents on metallic products), ASTM, ASME (American Society of Mechanical Engineers), International Organization for Standardization (ISO), CARES (Certification Authority for Reinforcing Steels), and a long list of customer-specific requirements all demand evidence that materials comply with defined specifications.

The MTR provides that evidence. Auditors, inspectors, certification bodies, and customers routinely request material documentation during quality reviews and compliance assessments.

Organisations that cannot produce valid certification records face:

• Audit findings
• Delayed approvals
• Customer disputes
• Contractual issues
• Regulatory penalties

The MTR is documented proof that you supplied and verified the correct material. In many cases, no certificate means no compliance.

The Challenge of Managing MTRs

Despite their importance, many organisations still manage Mill Certificates by hand. Documents end up scattered across:

• Email folders
• Shared drives
• Filing cabinets
• Spreadsheet trackers

As certificate volumes grow, the challenges grow with them. Quality teams burn hours on:

• Searching for certificates
• Verifying heat numbers
• Matching documentation to inventory
• Building customer quality packs
• Preparing for audits

This piles on administrative burden and increases the risk of errors. The information exists, but it stays trapped inside thousands of PDF documents.

Turning MTRs into Business Intelligence

Forward-thinking manufacturers are changing how they handle certification data. Instead of treating MTRs as documents, they treat them as valuable business assets.

When you digitise and structure certificate data, you gain the ability to:

• Search materials instantly
• Improve inventory traceability
• Monitor supplier quality
• Accelerate audits
• Build compliance packs automatically
• Answer customer documentation requests in minutes

What was once an administrative task becomes a source of operational intelligence.

How GoSmarter Mill Cert Manager Helps

GoSmarter Mill Cert Manager transforms Mill Certificates from static PDFs into searchable, structured data. Using AI, the platform automatically extracts:

• Heat numbers
• Material grades
• Dimensions
• Chemical properties
• Mechanical properties
• Certification information

This lets manufacturers, fabricators, stockholders, and distributors build a complete digital traceability system without manual data entry. Teams locate certificates instantly, verify material information, and keep audit-ready records at all times.

The result is stronger compliance, improved traceability, and significantly less administrative effort.

More Than a Piece of Paper

A Material Test Report can look like just another document in the procurement process. In reality, it is one of the most valuable assets in the metals supply chain.

It proves quality. It enables traceability. It supports compliance. Managed well, it gives you the confidence to operate safely, efficiently, and competitively.

In metals manufacturing, compliance starts with documentation. Operational excellence starts with knowing exactly where your material came from and where it is going.

Less Paper. More Metal. GoSmarter Mill Cert Manager helps manufacturers automate certificate management, strengthen traceability, and stay audit-ready every day.

Frequently Asked Questions

Further Reading

• GoSmarter Mill Cert Manager — automatically read mill certs and stop typing data
• AI for Mill Test Report Traceability
• Audit Trails: Because ‘I Think It Was Dave’ Isn’t a Defence
• How to Automate Mill Certificate Management in 5 Steps
• Integrated Cert Traceability hub

One document holds that story: the Mill Test Certificate (MTC), often called a Material Test Report (MTR).

Most procurement teams treat Mill Certs as paperwork. The certificate arrives with the delivery, someone files it for compliance, and everyone moves on. The best manufacturers, fabricators, and stockholders see it differently. Your Mill Certs are your superpower. They are the foundation for traceability, quality assurance, risk management, customer confidence, and operational excellence.

What is traceability?

Traceability is the ability to track material across its entire lifecycle. It follows the metal from the producing mill to the finished product you deliver to the customer.

A strong traceability system answers critical questions instantly:

• Where did this material originate?
• Which heat or batch did it come from?
• What chemical and mechanical properties does it have?
• Which supplier provided it?
• Which customer received it?
• Which standards and specifications does it meet?

In regulated industries, these questions are not optional. They are fundamental. Traceability protects safety, quality, and compliance. It applies whether you make structural steel, pressure vessels, offshore components, rail infrastructure, aerospace parts, or medical devices.

Why traceability matters more than ever

Modern supply chains are more complex than ever. Material often passes through several suppliers, processors, distributors, fabricators, and subcontractors before it reaches its final home. Every handoff creates risk.

Without proper traceability, you expose your business to material mix-ups, compliance failures, and quality disputes. Product recalls, customer claims, regulatory penalties, and reputational damage follow.

When something goes wrong, you need to identify the affected material quickly and confidently. The companies that do this well protect both their customers and their bottom line.

Mill Certs: the backbone of traceability

A Mill Cert is documented proof that material meets the required specification. It carries the critical detail:

• Heat numbers
• Material grades
• Dimensions
• Chemical composition
• Mechanical properties
• Applicable standards
• Manufacturer information
• Testing results

Each certificate creates a permanent link between the physical material and its manufacturing origin. Break that link and true traceability becomes impossible.

Every bar, plate, beam, coil, pipe, or fitting traces back to a specific heat and production batch via the certificate. That is why Mill Certs are often the most important documents in a metals business.

The hidden value of heat numbers

A heat number is far more than a reference code. It acts as the DNA of a material batch.

Say a quality issue surfaces months, even years, later. The heat number lets you:

• identify affected material immediately
• locate stock still in inventory
• work out which customers received material from the same batch
• isolate the risk
• prevent wider disruption

Instead of investigating thousands of tonnes, you focus on the specific batches affected. What could have become a major crisis becomes a controlled quality process.

Traceability creates competitive advantage

Many companies treat traceability as a compliance requirement. Leading organisations treat it as a competitive advantage.

Customers increasingly demand proof of quality, sustainability, compliance, and origin. When a customer requests certification documentation, a business with strong traceability responds in minutes, not hours or days.

The payoff shows up across the operation:

• Faster project approvals
• Greater customer confidence
• Less administration
• Better audit readiness
• Stronger supplier relationships
• Higher quality standards

In competitive markets, these advantages matter.

The problem with manual traceability

Despite all this, many businesses still manage Mill Certs by hand. Certificates sit in email folders, on shared drives, renamed manually, linked to spreadsheets, or filed in physical folders.

This approach creates real problems. Finding a certificate takes hours. Building a customer quality pack becomes a manual chore. Heat-code searches turn painful. Audit preparation gets stressful.

Worst of all, the traceability information stays trapped inside portable document format (PDF) files. It never becomes searchable business data.

Turn Mill Certs into a strategic asset

The most forward-thinking manufacturers turn their Mill Certs from static documents into digital assets. When you digitise and extract certificate information, you build a searchable traceability system. It connects procurement, inventory, production, quality, compliance, and customer deliveries.

Instead of searching through folders, your team locates any material record, heat number, grade, supplier, or certification history in seconds. That changes the role of the Mill Cert entirely. It becomes operational intelligence, not administrative paperwork.

How Mill Cert Manager strengthens traceability

Mill Cert Manager (built by Nightingale HQ) was made to solve exactly this challenge for metals manufacturers, fabricators, stockholders, and distributors.

Using Artificial Intelligence (AI), the platform automatically extracts and structures the key information from a Mill Cert:

• Heat numbers
• Material grades
• Dimensions
• Chemical properties
• Mechanical properties
• Standards and certifications

The result is a digital traceability system. You search certificates instantly, track material across the supply chain, and build compliance packs quickly. Audit readiness improves, manual admin drops, and customer confidence grows. Most importantly, your critical material information stays accessible when it matters most.

When Mill Cert Manager feeds clean data into Metals Manager, the link between cert, stock, and order never breaks. You get integrated cert traceability from goods-in through to despatch.

Your Mill Certs are more than compliance documents

Every delivery of steel, aluminium, stainless steel, titanium, or speciality alloy arrives with valuable data attached. The only question is whether that data stays buried in PDFs or becomes a strategic asset.

The companies that embrace traceability do more than meet a compliance requirement. They build stronger supply chains, improve quality, reduce risk, and create competitive advantage.

In modern metals procurement, traceability is not just about knowing where material came from. It is about knowing exactly where your business is going.

Less paper. More metal. With Mill Cert Manager, your Mill Certs become more than documents. They become your superpower.

Frequently Asked Questions

Further Reading

• Mill Cert Manager: Automate Mill Certificate Processing — extract and link your incoming Mill Certs automatically
• AI for Mill Test Report Traceability
• Audit Trails: Because ‘I Think It Was Dave’ Isn’t a Defence
• Metals Manufacturing Glossary — heat numbers, MTCs, and traceability in plain English

MTR compliance is not just a quality box to tick if you buy for manufacturing, construction, energy, oil and gas, aerospace, automotive, or infrastructure. It is a business necessity. The logic is simple. If you cannot verify your materials, you cannot guarantee your products.

“Confidence starts with traceability. And traceability starts with the Mill Certificate.”

Why MTR Compliance Matters

Every shipment of metal arrives with a promise. The supplier promises the material matches the order, complies with the relevant standards, and suits its intended use. The MTR is the evidence behind that promise.

A valid MTR documents:

• Material grade and specification
• Heat number and batch identification
• Chemical composition
• Mechanical properties
• Testing results
• Applicable industry standards
• Manufacturer certification

Without a valid MTR, you have no reliable way to confirm the material meets the specification. Where safety, quality, and traceability are critical, that gap creates real risk.

Not All MTRs Are Created Equal

One common mistake is assuming every Mill Certificate carries the same level of detail. It doesn’t. Requirements vary by industry sector, customer requirement, product application, regulatory obligation, and material standard.

A structural steel project needs different certification evidence than an offshore energy project. A pressure vessel manufacturer needs more detailed verification than a general fabricator. Work out what level of certification you need before you buy. Skip that step and you risk rejected deliveries, delayed projects, or expensive rework.

The Buyer’s Responsibility

Many teams assume compliance is the supplier’s job alone. It isn’t. Suppliers provide the documentation. Buyers are responsible for verifying it meets their quality and compliance requirements.

Every procurement process should include:

• Verification of received MTRs
• Validation against purchase order specifications
• Confirmation of applicable standards
• Traceability checks
• Record retention procedures

The best procurement teams treat MTR verification as a standard part of goods receipt and quality control.

Build a Strong MTR Compliance Process

Good compliance starts before the material arrives. Set clear expectations with suppliers and require:

• Mill Certificates for every batch supplied
• Digital copies before shipment where possible
• Full traceability information
• Compliance with the specified standards
• Consistent document formats

This cuts surprises and makes the whole process more efficient. The rule is simple: no certificate, no acceptance.

The Essential MTR Audit Checklist

Every buyer needs a standard process for reviewing Mill Certificates. Before material enters inventory, verify these seven things.

1. Heat number verification

The heat number is the foundation of traceability. Confirm that it appears on the certificate, matches the material markings, and aligns with your inventory records. Without a valid heat number, you lose traceability from the start.

2. Material grade confirmation

Verify the certified grade matches the purchase order, customer specifications, engineering drawings, and project standards. Even small discrepancies create compliance issues later.

3. Chemical properties review

Check the chemical composition sits within the limits for the material standard. Pay particular attention to carbon content, manganese, sulphur, phosphorus, and alloying elements. For welding applications, verify the carbon equivalent value too.

4. Mechanical properties validation

Review yield strength, tensile strength, and elongation. Check impact values and hardness results where the standard specifies them. Confirm every result meets the required standard.

5. Certification statements

Verify the document clearly states compliance with the relevant standard, specification, or purchase order. Common examples include EN 10204 3.1, applicable material or testing standards, and customer-specific specifications.

6. Laboratory test results

Review the supporting test results and confirm they relate directly to the certified heat or batch. Testing information must be complete, legible, and relevant to the material supplied.

7. Inventory reconciliation

Confirm quantities match the delivery records, descriptions are consistent, and the traceability data links to your inventory system. This keeps traceability intact throughout the material lifecycle.

Industry-Specific Compliance Requirements

Different industries demand different things from material certification.

The Hidden Challenge: Managing Thousands of Certificates

As an organisation grows, managing Mill Certificates gets harder. Many businesses still rely on shared drives, email folders, physical filing systems, and spreadsheets.

The result is predictable: lost certificates, time-consuming searches, manual data entry, audit preparation headaches, and rising compliance risk. When an auditor or customer asks for a specific certificate, finding it becomes a project in itself.

Modernise MTR Compliance with GoSmarter

The most effective procurement teams have moved beyond manual certificate management. GoSmarter MillCert Reader uses Artificial Intelligence (AI) to extract, classify, organise, and manage Mill Certificates automatically.

Instead of spending hours reviewing certificate files, your team can:

• Search certificates instantly
• Validate heat numbers in seconds
• Access material properties on demand
• Build compliance packs quickly
• Improve traceability across the supply chain
• Keep audit-ready records every day

The result is a procurement process that is faster, more accurate, and far more compliant.

Compliance Is More Than a Tick-Box Exercise

People often treat Mill Certificates as paperwork. In reality, they are one of the most important risk-management tools a procurement team has. A strong MTR compliance process protects product quality, customer relationships, regulatory compliance, brand reputation, and business continuity.

Treat Mill Certificates as strategic assets rather than admin, and you gain a real competitive advantage. In metals procurement, confidence starts with traceability. And traceability starts with the Mill Certificate.

Less paper. More metal.

Frequently Asked Questions

Further Reading

• GoSmarter MillCert Reader — automate extraction and verification of incoming mill certificates
• GoSmarter Compliance Solution — map your certification process to the standards that apply to you
• Audit Trails: Why They Matter in Metals Manufacturing
• How to Automate Mill Certificate Management in 5 Steps
• Case Study: MillCert Reader Saves 10 Hours a Month
• Metals Manufacturing Glossary — definitions for MTR, heat number, EN 10204, and more

They appear on Mill Test Certificates (MTCs), Material Test Reports (MTRs), purchase specifications, engineering drawings, and quality documentation across the world.

But what exactly are ASTM standards, and why do they matter so much to manufacturers?

What Are ASTM Standards?

ASTM International (formerly the American Society for Testing and Materials) is one of the world’s largest standards development organisations. Founded more than 125 years ago, ASTM develops internationally recognised standards covering materials, products, testing methods, and manufacturing processes. Today, more than 12,000 ASTM standards are used globally. They span construction, manufacturing, aerospace, and energy.

These standards provide a common language. Manufacturers, suppliers, customers, and regulators all work to the same specifications, regardless of location.

Put simply, ASTM standards help ensure that a product made in one country performs exactly as expected when used in another.

Why ASTM Standards Matter

Manufacturers face growing pressure to prove quality, consistency, and compliance.

Without recognised standards, every supplier would define materials differently. Procurement, quality control, and international trade would become far more complex.

ASTM standards help organisations:

• Improve product quality
• Ensure material consistency
• Reduce manufacturing risks
• Simplify supplier qualification
• Support regulatory compliance
• Enable international trade
• Improve customer confidence

For manufacturers in regulated sectors, ASTM standards are often the backbone of the quality assurance system.

ASTM and Metals Manufacturing

In the metals industry, ASTM standards define many of the materials used every day.

Examples include:

• ASTM A36 — Structural Steel
• ASTM A516 — Pressure Vessel Plate
• ASTM A240 — Stainless Steel Plate
• ASTM A312 — Stainless Steel Pipe
• ASTM F3125 — Structural Bolts

These standards specify requirements such as:

• Chemical composition
• Mechanical properties
• Manufacturing requirements
• Testing procedures
• Inspection criteria
• Acceptance limits

When a mill certificate references an ASTM grade, it confirms that the material was produced and tested against those defined requirements.

Different Types of ASTM Standards

Many people think ASTM only covers material grades. The standards are much broader.

Material Specifications

These standards define what a material must contain and how it should perform. Examples include steel, aluminium, fasteners, plastics, and composites.

Test Methods

These standards define how testing is carried out, so results stay consistent and repeatable.

Examples include:

• Tensile testing
• Hardness testing
• Impact testing
• Chemical analysis
• Wear and abrasion testing

A test performed in Cardiff should produce comparable results to the same test performed in Chicago, Mumbai, or Düsseldorf.

Practices and Guides

These provide recommended approaches for inspection, quality management, and technical procedures.

Terminology Standards

These ensure that technical terms carry consistent meanings across industries and regions.

ASTM and International Trade

Although ASTM originated in the United States, it has become a truly global standards organisation.

Manufacturers in more than 100 countries rely on ASTM standards to source materials, qualify suppliers, and confirm compliance. ASTM’s membership includes technical experts from over 140 countries who develop and maintain the standards.

For companies exporting products internationally, ASTM standards often act as a passport. They help materials and products move more easily through global supply chains.

ASTM Standards and Mill Certificates

For quality managers, procurement teams, and auditors, ASTM standards are most visible on Material Test Reports and Mill Test Certificates.

Every day, manufacturers manually review certificates to confirm:

• Material grades
• Heat numbers
• Chemical composition
• Mechanical properties
• Compliance with customer specifications

The challenge is that these documents arrive in different formats from different suppliers.

A large manufacturer may process thousands of certificates every year.

The Growing Compliance Challenge

As supply chains grow more complex, managing ASTM-related documentation becomes harder.

Common challenges include:

• Missing certificates
• Inconsistent document formats
• Manual data entry
• Time-consuming audits
• Traceability gaps
• Supplier compliance verification

Many organisations still rely on spreadsheets, email folders, and shared drives to manage critical compliance information.

The result is significant administrative effort and increased risk.

How Artificial Intelligence (AI) Is Transforming ASTM Compliance

Forward-thinking manufacturers now use AI to automate certificate processing and compliance management.

Instead of reviewing every document by hand, AI can:

• Extract ASTM grades automatically
• Identify heat numbers and material properties
• Validate certificate data
• Detect missing information
• Create searchable digital certificate libraries
• Support audit preparation
• Improve material traceability

This cuts administration while improving visibility, consistency, and compliance across the organisation.

GoSmarter’s MillCert Reader, built by Nightingale HQ, does exactly this for ASTM and EN 10204 certificates. It reads the grade, heat number, and measured properties from each document. It then validates them against the expected ranges for that grade. Non-conforming material is caught at goods-in, not halfway through a job.

Less Paper. More Metal.

ASTM standards are far more than technical references buried in engineering specifications.

They provide the framework that lets manufacturers around the world produce, test, verify, and trade materials with confidence.

For quality and compliance teams, the challenge is not understanding ASTM standards. It is managing the growing volume of documentation that comes with them.

That is why manufacturers now use AI-powered tools to automate certificate management, improve traceability, and cut hours of manual admin.

Because every hour spent chasing paperwork is an hour not spent making product.

Upload your first cert — it takes ten seconds →

Or see what GoSmarter does with your actual certificates. Bring one along to the call.

Frequently Asked Questions

Go Deeper

• Mill Certificate Automation for Metals Manufacturers — the complete guide to what GoSmarter does with mill certs
• GoSmarter MillCert Reader product page — features, pricing, and free trial
• AI for Mill Test Report Traceability — how AI cuts MTR handling from minutes to seconds
• BS EN 1090 and NSSS Compliance — the structural steel paperwork standards demand

GoSmarter is made by Nightingale HQ, a UK-based AI company building practical tools for metals manufacturers since 2018.

It’s a trick question. EN 10204, ASTM, and ASME are not rival standards. They answer three different questions, and a single certificate often satisfies all three at once.

Here’s how they fit together.

Three Standards, Three Questions

The quickest way to keep them straight:

• EN 10204 — is an inspection document standard — it answers “what kind of certificate is this and who authorised it?
• ASTM — is a material specification — it answers “what are the chemical composition, tensile strength, and impact requirements this steel must meet?
• ASME — * it answers “is this material approved for use in ASME-coded pressure equipment?“How can this material be used safely?*

EN 10204 governs the document. ASTM governs the material specification. ASME governs the equipment built from it. They sit on different axes, so they don’t compete but they stack.

EN 10204: The Certificate Type

EN 10204 is the European standard that defines the type of inspection document a mill issues. It says nothing about what the steel is. It only says how the test results were produced and who signed them off.

There are four types:

When a customer asks for “3.1 certs,” this is the standard they mean. They are specifying the level of assurance, not the material grade. Supplying a 2.2 when the purchase order says 3.1 is a non-conformance, even if the steel itself is perfect.

ASTM: The Material Specification

ASTM (the American Society for Testing and Materials, now ASTM International) develops the specifications that define a material. An ASTM designation tells you the chemical composition, the mechanical properties, and the tests the metal must pass.

You will recognise a few:

• ASTM A36 — structural steel
• ASTM A516 — pressure vessel plate
• ASTM A312 — stainless steel pipe

So ASTM answers “what is this material?” Europe answers the same question with its own EN material specs: EN 10025 for structural steel, EN 10028 for pressure vessel plate. ASTM and EN 10025 sit on the same axis. EN 10204 sits on a different one.

ASME: The Design Code

The American Society of Mechanical Engineers (ASME) develops the codes that govern how equipment is designed, fabricated, inspected, and operated safely. The best known is the ASME Boiler and Pressure Vessel Code (BPVC).

ASME doesn’t define the steel. It defines the rules for building safe equipment from it. When a material is adopted into the BPVC for code work, it gets an “SA” prefix. ASME SA516 is the code-adopted version of ASTM A516. The chemistry and properties are virtually identical. The SA designation just means it’s cleared for use in code-compliant pressure equipment.

How They Stack on One Certificate

Here’s the part that confuses people. A single mill certificate can carry all three at once.

Picture a plate of pressure vessel steel:

• The material is certified to ASTM A516 Grade 70 (or its dual designation, ASME SA516 Grade 70).
• The certificate is issued as EN 10204 Type 3.1 — the manufacturer’s representative has signed off the test results for that specific heat number.
• The finished vessel is then fabricated to ASME Section VIII.

Nothing here is in conflict. EN 10204 describes the paperwork. ASTM/ASME describe the material and the equipment. One document, three boxes ticked.

What to Check on a Purchase Order

Because these are separate axes, a purchase order needs to pin down all of them. Vague specs cause rejected material and delayed shipments. Check:

• Certificate type — is it 3.1, 3.2, or will 2.2 do?
• Material specification — ASTM A516, EN 10028, or both (dual-certified)?
• Design code, where it applies — ASME Section VIII, EN 13445, or a client spec?

Get one wrong and you have a non-conformance, even when the other two are correct.

Why UK and EU Manufacturers See American Standards

If you fabricate in Britain, why are ASTM and ASME on your desk at all? Global supply chains. Plenty of UK and EU work is built to American codes. Export contracts, oil and gas projects, and multinational clients who standardise on ASME all drive that. Mills routinely dual-certify plate to both ASTM and the equivalent EN spec so it can be sold either way.

The upshot: your quality team has to read certificates against more than one framework, often on the same project.

Why This Matters for Traceability

For procurement, quality, and welding inspection, mixing up these standards is expensive. The risks are familiar:

• Wrong certificate type accepted against a 3.1 requirement
• Material that meets EN but not the specified ASTM grade
• Failed audits when the chain of evidence doesn’t line up
• Rejected material, rework, and slipped deadlines

When you review certificates, you need confidence on all three axes: the right document type, the right material spec, and any design-code requirements. That gets hard fast when you’re working across hundreds of certificates and dozens of suppliers.

Less Paper. More Metal.

Most teams still check this by hand, it’s often done by opening PDFs one by one, hunting for the cert type, the grade, the heat number, and matching each against the order. It’s slow and easy to get wrong.

GoSmarter’s MillCert Reader, built by Nightingale HQ, reads the certificate and pulls out the EN 10204 type, the ASTM or ASME grade, and the heat number. It checks the values and files everything so you can find it in seconds. The result is faster compliance, cleaner traceability, and far less admin.

EN 10204, ASTM, and ASME aren’t competing. They describe the document, the material, and the equipment. Once you see them as three questions rather than three rivals, the certificate stops being confusing.

Because every hour spent decoding a certificate is an hour not spent making metal.

Frequently Asked Questions

Go Deeper

• ASTM vs ASME: What’s the Difference?
• AI for Mill Test Report Traceability
• EN 10204 and the four certificate types

The fix is straightforward: lock down network zones, remote access, model changes, device data and backups. This matters for production managers, engineers and IT teams running tools like GoSmarter Product Lineage, Business Manager, and Production Planner across metals manufacturing.

What you get from this guide:

• The main risks, without the sci-fi nonsense
• Where plants usually get caught out, from IT-to-OT movement to poisoned data
• What to fix first, so you don’t waste time on theatre
• How to keep the line moving if systems go down

Here’s how to fix it.

Industrial Controls Are a Hacker’s Dream

Before we dive in, take a look at this video:

Lock Down the Network Before Attackers Walk In

Seventy-five per cent of operational technology (OT) attacks start on the IT side. Usually it’s something boring and avoidable, like a phishing email or a compromised enterprise resource planning (ERP) server [3]. Once attackers get a foothold, a badly segmented plant network does the rest of the work for them.

1. Split OT, IT and AI into Separate Zones

Keep business systems, plant systems and AI systems in separate zones. Put a broker or demilitarised zone (DMZ) between them. If ransomware hits your email server, it should stop there, not drift into the AI model running your production schedule.

Build layered zones that split business systems, plant systems and physical controls. Keep AI servers in their own plant zone. Don’t let them share a network segment with your ERP or anything internet-facing. A controlled broker between ERP and plant-floor AI gives you a defined conduit for data, not a wide-open pipe [2][3].

For one-way data flows, such as sensor readings, camera feeds and historian data leaving the shop floor for a cloud AI model, use data diodes. They physically block any return traffic from reaching the production floor [7].

It’s a hardware guarantee, not a firewall rule someone can misconfigure.

2. Tighten Remote Access and Cloud Connections

Zones help, but remote access can still punch straight through them. Every remote session is a risk. That includes a maintenance engineer checking an AI dashboard from home or a supplier looking at a quality inspection feed.

• Require multi-factor authentication (MFA) on every remote connection. It cuts 91% of credential-based breaches in industrial settings [4].
• Use on-demand virtual private networks (VPNs) and approved jump hosts so external parties never connect straight to plant systems.
• Block direct vendor access to OT and AI systems.
• Rotate all service-account credentials and application programming interface (API) keys used for SCADA (Supervisory Control and Data Acquisition), historian and AI links every 90 days [2].

You also need a fast kill switch for outside access. If you can’t isolate a vendor session within minutes, you’re not containing anything. You’re just watching the breach move.

3. Sort Out Patching and Basic Hardening on Every Device That Touches AI

Every device that feeds data into, or takes instructions from, an AI system belongs on your patching register. That means edge gateways, vision system servers, historian boxes, the lot.

Legacy programmable logic controllers (PLCs) are often the awkward bit. If they can’t support encryption natively, use secure gateways as encryption proxies. They can talk Modbus locally, then tunnel data over TLS (Transport Layer Security) 1.3 to the rest of the network [4]. Set firewall allowlist rules so AI nodes can talk only to named historian or management plane IP addresses [2]. Not the whole plant subnet.

Disable unused ports and services on every edge box and camera. If a device touches your AI pipeline, give it an owner and patch it.

Once the network is segmented and patched, the next job is stopping the wrong people from changing the model, the data and the dashboards.

Control Who Can Change Models, Data and Dashboards

A bad login can do as much damage as malware. If someone can quietly move a quality pass/fail threshold, edit scrap records or change scheduling rules, your network segmentation will not save you. Access control is where you stop that kind of damage.

4. Give People Only the Access They Need

Least privilege means each person and device gets only the access needed for the job.

The rule is simple: give every person and every device the minimum access needed for the task, and no more. Set up role-based access control, with clear tiers and clear limits.

“Role-Based Access Control (RBAC) ensures that a maintenance technician can view hydraulic health scores, but only a senior process engineer can authorize setpoint adjustments to the HAGC system.” - Alex Jordan, iFactory [4]

Use MFA for AI dashboards and admin consoles. For on-site access, use badge tap or biometrics as the second factor only. When someone leaves or changes role, remove access that same day. Better yet, automate it. Sync the AI platform with your central directory or ERP roles, so if an ERP account is revoked, AI access dies with it too [3].

Access is only part of the job. The records feeding the models need the same discipline.

5. Protect Training Data, Mill Certificates and Scrap Records

Training data matters as much as the model. Poisoned scrap records, altered mill certificates, or manipulated production parameters can skew AI outputs and burn cash fast.

Sensitive records need access rules, not just passwords. Lock down who can read, edit, or export training data, quality images, mill certificates, and scrap logs. Store them in encrypted, access-controlled systems. Restrict exports. Use named accounts. Log every read, edit, and download.

If your data is clean, the next problem is the model itself changing behind your back.

6. Treat Model Changes Like Tooling Changes, Not Casual Tweaks

Treat AI changes like tooling changes. That means formal approval, version control, and a rollback path.

Undocumented model changes can quietly push up scrap and rework before anyone notices the drift. That is the danger. Not some sci-fi machine takeover. Just small, sloppy changes that hit your margins.

Require dual sign-off for any retraining event or threshold change that affects safety, quality, or margin [2]. Keep a versioned model registry so you can roll back fast [2]. Use signed model files and hash checks for all model artefacts, so a tampered or poisoned model gets rejected before it reaches production [2][3]. Log every change, approval, and deployment in an immutable audit trail [4][3].

The goal is simple: no silent sabotage. If a model changes, someone approved it, it is logged, and you can prove it.

Even locked-down models will fail if the shop-floor data feeding them is spoofed or poisoned.

Stop Bad Device Data from Turning Your AI Against You

Locked-down models and tight access controls do nothing if the data coming off the shop floor is already rotten. A spoofed vibration sensor, a tampered camera feed, or a gateway set up wrong can shove your AI into bad decisions with no warning.

7. Know Every Camera, Sensor and Edge Box Feeding Your AI

If you haven’t counted the devices feeding your AI, you don’t control them. Start with a full device register. That means every defect-detection camera on the inspection line, every vibration sensor on a rolling mill bearing, and every edge box exporting SCADA tags to a scheduling or maintenance model. In one industrial site, asset mapping found 139 unknown entry points [7]. That’s 139 attack paths nobody even knew were there.

For each device, write down:

• what data it sends
• which system receives it
• who owns that connection
• whether the flow is one-way or two-way

Then deal with the obvious bit people skip. If data only needs to move outwards, such as sensor telemetry feeding an AI inference node, use hardware-enforced data diodes.

“Firewalls are bidirectional. Data diodes are not… A diode lets production data reach AI tools while physically preventing traffic from moving back into OT through the same path.” - OPSWAT Team [7]

Use automated discovery and passive monitoring. Standard scans miss rogue devices and config drift [12]. Lock control cabinets and access rooms so people can’t just wander up to edge devices and sensors [11]. Log every device enrolment, replacement and rekeying. Then block any telemetry that turns up without a verified device identity. If the endpoint isn’t verified and the telemetry isn’t signed, stop it before it gets anywhere near the AI pipeline [2][6].

Once the network is sorted, the next problem is the data itself.

8. Watch for Poisoned Data, Spoofed Signals and Odd AI Output

After you’ve counted every device, watch what it sends. Look for drift, spoofing and sudden change.

A clean device register helps, but it won’t save you from poisoned inputs. Attackers can alter sensor readings, corrupt quality images or, in metals manufacturing, stick physical adversarial patches on products to fool computer vision inspection systems into passing defective parts without touching the network at all [6].

So watch the data, not just the network. Set up real-time drift checks on model inputs. If vibration signatures, bearing temperatures or defect rates suddenly shift, that should fire an alert instead of quietly nudging the model off course [2][6]. A second model checking incoming sensor values before they hit the main AI gives you another layer of defence [11]. For computer vision, checking early vision-layer signals can catch odd patterns before they turn into a bad classification [6].

Use:

• drift detection for poisoned images
• TLS 1.3 and certificate checks for spoofed sensor data
• hash checks for tampered model files

Manual checks won’t cover this. Automated monitoring won’t catch every trick either, but it cuts the time between attack and detection by a lot.

9. Keep a Human in the Loop When the Call Could Cost Real Money

Data can look clean and still be wrong. That’s why automation should handle routine calls only. If an AI recommendation could trigger a maintenance shutdown, change a production schedule, or pass a batch of metal that may be out of spec, a person needs to sign it off.

This isn’t about slowing the plant down for the sake of it. It stops one bad output from a spoofed sensor or poisoned feed turning into missed deliveries, excess scrap or unsafe product.

Ninety-one per cent of manufacturing security professionals said they need to understand how AI makes decisions before they trust it [5].

The rule is simple. Any AI action that touches safety interlocks, a critical quality pass/fail decision, or a scheduling change tied to a committed delivery date needs explicit human confirmation [6]. Put that approval step in a system the AI cannot touch. If the AI can approve its own recommendations, your human-in-the-loop setup is just theatre. If bad input still slips through, the next line of control is recovery: backups, restore steps and a fallback way to run the line.

Plan for the Day It All Goes Wrong

Once access and data controls are in place, recovery is your last line of defence. When ransomware hits a plant, average downtime reaches 21 days [8]. That’s not a rough week. That’s lost orders, missed delivery dates, and compliance mess that keeps biting long after the screens light back up.

10. Back Up Models, Configs and Integrations So You Can Rebuild Fast

Back up everything you need to rebuild fast and clean. That means a signed, hashed model registry, so you can check the restored version hasn’t been tampered with [2]. It also means backing up PLC programmes, human-machine interface (HMI) images, process recipes, and the integration mappings between your manufacturing execution system (MES), ERP, and shop-floor systems. Store them offline or in immutable storage [10].

For GoSmarter users, this is pretty plain:

• Your Product Lineage setup links inventory to heat codes and lets you pull mill certificate PDFs by heat code.
• Your Business Manager setup holds customer, supplier, and order data.
• Your Production Planner setup, along with the current order backlog, should sit in the backup set too.

Aim to roll back to a verified AI model state within 15 minutes [2]. Then test it every quarter.

Backups are just theory until you can prove the restore works.

11. Write the Fallback Plan Before Ransomware Writes It for You

In August 2025, Jaguar Land Rover suffered a total production halt across three UK plants after attackers used stolen Jira credentials to move from corporate IT into production networks. The forensic investigation took five weeks to verify every system before restart, and the incident cost the UK economy an estimated £1.9 billion and required a £1.5 billion government-backed loan to keep the supply chain solvent [1].

If recovery takes hours, you still bleed cash. The bigger risk isn’t AI failure. It’s not knowing your fallback when the main system falls over.

Your fallback plan needs named owners and plain decisions:

• Who calls the halt to production
• How scheduling drops back to a manual board or offline spreadsheet
• How quality inspection carries on with trained operators and manual gauge control

Keep the last clean Business Manager export. Keep paper processes for scrap records, heat code tracking, and product lineage too. Run a tabletop drill once or twice a year [10][9]. Walk the night shift through a false 0% vibration reading or a vision system that misses known defects [9]. Better to find the gaps now than in the middle of a plant shutdown.

Run a GoSmarter Access and Backup Check This Week

Do this this week. Give one person an hour and check three things.

First, who has access to your GoSmarter account, and does each person still need it? Remove anyone who has left or changed role.

Second, are Product Lineage, Business Manager, and Production Planner reachable only from the zones they should be in?

Third, are your configurations backed up offline, and have you tested the restore?

If the answer to any of those is “not sure,” start there. One hour this week is a lot cheaper than five weeks offline.

FAQs