How facilitation turns AI ideas into AI products

The three assumptions holding manufacturing companies back

Most established manufacturing companies fall into one of three traps when it comes to AI.

The first: "It doesn't really affect us." The second: "We've rolled out Microsoft Copilot, so our AI strategy is sorted." The third: "Our processes are too specific, too physical, too human. AI can't touch them."

All three are understandable. But all three are wrong.

And here's the hard truth: companies that cling to these assumptions don't just stand still. They fall behind. Quietly, gradually, and then suddenly.

But before you commission a six-figure AI project or hire a team of developers, there's one step almost every company skips. A step that decides whether an AI initiative succeeds or turns into an expensive lesson that poisons the well for years to come.

That step is facilitation.

What facilitation really means in an AI context

Facilitation isn't a buzzword here. It's not a workshop where someone draws on a whiteboard and everyone nods politely. It's a structured process that brings your department heads into one room and creates space for something rare: an honest conversation about what's actually broken, what's wasting time, and where smarter decisions could get made faster.

The goal isn't a long list of AI ideas. It's finding the ideas worth pursuing, the low-effort, high-impact opportunities hiding in your existing processes. The ones your people already know about but haven't had the right forum to raise.

Sounds simple. It isn't. Because the moment you put a group of department heads in a room to talk about AI, you'll run into two very specific personality types. And if you don't handle them carefully, they'll derail the whole thing.

The sceptic and the evangelist

You know these people. Maybe you're one of them.

The sceptic has watched technology initiatives come and go. They've seen ERP rollouts blow the budget, CRM systems gather dust, and automation projects promise the world and deliver nothing but headaches. When someone says "AI", they hear "cost." They'll tell you, with total conviction, that AI can't do what your business needs, that it's a distraction from Silicon Valley, and that the money would be better spent on new machinery.

The evangelist is the opposite problem. Their answer to every challenge is "AI will fix that." A quality control problem? AI. Supply chain complexity? AI. Recruiting? AI. They've read the articles, watched the keynotes, and are ready to automate everything right now.

Both have an important role to play. But left unguided, the sceptic kills momentum and the evangelist chases a mirage.

The way to get both onto productive ground is the same: real examples. Not chatbots. Not the generic AI assistant sitting in the browser bar. AI that works quietly in the background: enriching CRM data automatically before a sales call, analysing images from the production line to flag quality issues before they become customer complaints, monitoring systems and alerting the right person at the right moment.

When people see what AI actually does in practice, the mood in the room shifts. The sceptic realises AI only works when it's specific, and that suddenly looks doable. The evangelist realises AI needs precise instructions and can't perform magic, and their ideas get sharper. That's when the real work starts.

The moment it becomes real

In almost every well-run workshop, there's a moment when you can feel the room change.

It happens when the discussion moves from vague to precise. From "AI could look at our quality images" to "AI analyses an image automatically uploaded by our inspection system, sorts it into one of our predefined defect categories, and flags it for review."

That sentence, that level of precision, does something to the people in the room. The sceptic stops arguing because they can see exactly what it would do. The enthusiast stops waving their hands because they know exactly what it takes. Everyone starts to understand that the path from a half-formed idea to a working tool isn't mystical. It's a process. And it starts with getting clear on the problem.

That's what a good facilitator produces: not a wish list of AI ideas, but a handful of clearly defined, genuinely testable concepts, rooted in your field, your data, and your people.

Your domain knowledge is the real competitive advantage

Most technology vendors won't tell you this: the AI itself isn't your competitive advantage.

Anyone can use Claude, ChatGPT, or Gemini. Anyone can bolt on an off-the-shelf chatbot. Anyone can subscribe to a SaaS tool with built-in AI features. That's just standard now.

What can't be replicated is decades of production know-how. Understanding why a certain reading at one point in the process signals a problem three steps downstream. The institutional knowledge of which customers need what kind of communication at which stage of an order. The hard-earned expertise sitting in the heads of your most experienced people, expertise that until now has been almost impossible to code into software.

That's the gold. And it only comes out when you give people the right space to talk about it.

Off-the-shelf AI gives you a tool. Facilitated, domain-specific AI gives you a capability your competitors can't easily copy.

Why AI projects fail after the idea is validated

Let's be honest about what comes next, because this is where a lot of companies fall apart.

AI projects aren't software projects in the traditional sense. You can't plan them like a waterfall. You can't lock scope in week one and expect to deliver it unchanged in week twelve. The teams building them learn as they go. Data everyone assumed was clean and integrated turns out to be fragmented, inconsistent, or just missing. A set of instructions that works perfectly in one context falls apart in another. You build something, test it, and sometimes start over.

And the costs catch people off guard. Not just developer time, which for someone with real AI expertise can easily run €150 an hour or more, but also the infrastructure, the API usage billed by token, and iteration cycles that don't fit a traditional project budget.

None of that is a reason not to do it. But it is a reason to think strategically about which problem you solve first.

If your company's first AI initiative doesn't deliver visible, everyday value for people doing real work, if it solves a problem nobody cares about or one that was never properly defined, you haven't just wasted money. You've made every future AI conversation harder. A failed project can set a company back years.

That's exactly why the work done before a single line of code gets written matters so much. A proper workshop, a vetted idea, and a proof of concept built fast and honestly: none of that is a luxury. It's protection. Protection for your budget, your team's trust, and your company's willingness to keep going.

Talk to your people before you talk to a developer

If you've got an AI idea and don't know where to start, start here: talk to the people who actually do the work.

Tell your department heads what you're thinking. Expect them to pick your idea apart. That's the point. That friction is the most valuable thing they can give you. It either sharpens the idea into something workable, or kills it early, before it costs you real money.

The worst thing you can do is fall in love with an idea, commission expensive development, and find out six months later that the problem you wanted to solve wasn't actually a problem, or that the people who were supposed to use the solution had a simpler, better approach in mind the whole time.

Involving your people isn't just good practice. In an AI context, it's how you get the expertise into the product. That's what makes it useful. It's the only way it sticks.

What this looks like in practice

A good AI workshop doesn't arrive with a pre-set answer. It creates the conditions for the right answer to emerge from your own organisation.

That means everyone gets a say, including the sceptics, especially the sceptics. It means mapping processes honestly, not as wishful thinking. It means looking for where three things intersect: what causes daily friction, what data you already have access to, and where a precise AI application could actually make a difference.

It means leaving the workshop with something concrete enough to build, and honest enough to be worth building.

One final word

If anything in this post hit home, if you recognise the traps, the personality types, or the half-formed idea you've been carrying around for months, you're already ahead of most people.

The next step isn't a purchase decision. It's a conversation.

A good facilitator won't push a product on you. They won't tell you AI is the answer before they've understood the question. And they'll be honest with you if your idea doesn't fit, because a project that shouldn't get built costs more than a conversation that says so.

If you think you've got something worth developing, don't wait until it's perfectly polished. Bring it, unfinished, into a room with the right people.

That's where the real work begins.

Back to overview