Looking to the future, digital tools and smarter ways of working are reshaping shopfloor performance. In this guest insight from Process Management International (PMI), we explore how practical innovation and industry expertise are helping manufacturers turn data into action, and resolving nonsensical generative output, to solve problems faster.
There’s a particular kind of frustration that experienced improvement teams recognise instantly
You have the right people in the room, you’re using a recognised method, and the discussion is focused and disciplined. Everyone understands the problem and its constraints. Yet, despite doing everything “properly”, the conversation begins to stall and the options on the table start to feel like variations on things that have already been tried.
That was the situation one engineering team found themselves in while working with PMI on a supply‑chain issue that would be familiar to many SC21 organisations.
A practical problem, handled properly
A component was manufactured overseas and transported by ship, where exposure to the elements during transit led to repeat instances of weather damage. To prevent this, a protective coating was applied before shipping. The coating solved one problem, but created another. Once the part arrived on site, removing that coating proved slow, labour‑intensive and disruptive to flow.
The team approached the issue sensibly, using TRIZ (Theory of Inventive Problem Solving) to explore trade‑offs and inventive possibilities. The people involved were experienced, the method was sound, and the conversation was constructive. In other words, this was not a case of poor practice or lack of effort. Even so, the group reached a point where progress slowed and no clearly workable option emerged.
For organisations used to operating within SC21 frameworks, this moment is familiar. It’s not a failure of structure, but a recognition that complex, constrained problems don’t always yield easily to even the best methods.
Experimenting with AI, and learning its limitations quickly
As a test, the team ran the problem through a general AI tool. The output wasn’t nonsensical, but it was generic, loosely connected to the specifics of the situation, and noticeably inconsistent in how it applied the TRIZ logic the team had been working with. Even when explicitly prompted to follow the method, the model tended to drift.
That difference matters in aerospace and defence supply chains where interesting ideas are not enough. Suggestions have to respect constraints, systems, and discipline, otherwise they simply add noise. But rather than writing AI off as a distraction, the team took a more considered approach.
Designing AI to follow the method
Together, they created a simple AI agent instead of relying on a generic model. The agent was loaded with relevant TRIZ content and reference material and was given explicit instructions on how to structure the problem-solving conversation. Crucially, it was designed to pause at key points, check assumptions, and only move forward once the necessary information was in place. In short, it was asked to follow the process.
When the team reviewed the output, the difference was immediately apparent. Some suggestions were clearly impractical, which was expected. Others, however, were credible enough to warrant further exploration and testing, and importantly had not emerged during the original workshop discussion. What had changed was not the sophistication of the technology, but the way it had been constrained and guided.
Why this distinction matters in SC21 environments
There’s an important difference between asking AI for ideas and using it to support structured problem-solving. Generic AI models tend to prioritise breadth and pattern‑matching. A well‑designed agent can be made to operate within the same logic and discipline that experienced practitioners use, widening the option space without abandoning method.
For SC21 organisations, where consistency, traceability and confidence in outcomes matter as much as innovation, that distinction is critical.
Where automation makes this usable beyond a workshop
This is also where automation becomes central to the discussion. One of the risks with any promising improvement approach is that it remains tied to a moment, a project or a particularly capable individual. The longer‑term benefit comes when learning and logic are embedded into everyday ways of working.
Automation platforms play a quiet but important role here. Not as a headline technology, but as a way of connecting structured problem-solving to real workflows, follow‑up actions and outcomes. When judgement, assumptions, and decision logic are captured consistently, organisations are far less likely to revisit the same issues repeatedly under slightly different conditions. Learning is retained, not rediscovered.
In supply‑chain environments operating under pressure, it reduces reliance on a small number of experts and increases confidence that good thinking is being applied as standard.
Judgement still sits with people
None of this removes the need for human judgement. Someone still has to frame the problem, recognise what information is missing, and decide which options make sense in practice. AI and automation don’t replace experience, they support it.
What changes is access. Structured thinking that previously relied on a small number of highly experienced individuals can be made available more consistently across teams, without losing rigour. That is a capability conversation rather than a technology one.
A subtle shift in improvement roles
One of the quieter implications of this approach is how improvement roles begin to evolve. As methods, judgement and learning are made explicit and embedded into systems, less time is spent rediscovering fundamentals and more time is spent refining approaches, testing outcomes and learning what really works.
This is where AI and Automation Practitioner capability starts to make sense. Not purely as technical specialists, but as a bridge between improvement practice and operational execution. These practitioners understand process, people and change, and can design how structured thinking flows through an organisation safely and consistently.
A grounded view of AI in improvement
There’s nothing especially radical in this story. What it shows is that AI is most useful when it’s shaped by the same discipline, structure and judgement that experienced practitioners already rely on. On its own, AI is good at generating options. When it’s designed to follow method, it helps teams explore those options without losing rigour. Automation then allows that structured thinking to move beyond a single workshop, team or individual and into everyday work.
Together they become practical enablers rather than shortcuts, which is important for organisations operating within SC21 frameworks where the challenge is ensuring that disciplined thinking is applied consistently, under pressure, and across complex supply chains without increasing risk or variability.
This is why AI and Automation Practitioner capability is starting to matter more. Not as a new technical specialism, but as a way of bridging improvement practice and day‑to‑day operational reality. These practitioners focus on designing how structured thinking is embedded, followed and sustained, rather than rediscovered project by project.
Seen this way, AI becomes less about experimentation and more about reinforcement. It supports organisations in protecting what works, while making it easier for more people to apply proven methods with confidence. And that is a practical and credible way to strengthen improvement in demanding environments.
Want to explore further?
PMI offer training, podcasts and webinars throughout the year. Here are some over the next two months:
Visit their website: https://pmi.co.uk/webinars/
