The first AI-in-operations pilot is often easier than the rollout. A motivated team, a clear problem, direct support, short paths: that's how a lighthouse line comes together.
The second site is different. Different systems, different culture, different product mix, different leadership, different data quality. That's exactly where you find out whether the pilot just worked or whether it's scalable.
The pilot proves possibility
A lighthouse line answers one question: can this approach create value on a real problem?
That matters. But it isn't enough for scale. Plenty of companies have good pilots that never get past the first line. Not because the technology is bad, but because the operating model, governance and standards are missing.
The rollout proves organisation
Scaling needs answers to a different set of questions:
- Which data objects have to be standardised?
- Which local differences are allowed?
- Who owns the use case in the plant?
- Who decides when a process deviates?
- How do findings get turned into standards?
- How do you build the capability of an internal team?
Those are lean and leadership questions.
Don't copy, translate
The most common rollout mistake is copying. Whatever worked on line 1 gets pushed onto line 2 as a template. But the material flow, product mix, equipment history and team maturity are all different.
A playbook works better: which logic stays the same, what gets adapted locally, which roles have to be filled, which data is mandatory, which is optional?
Jidokai as a scalable way of working
In a rollout, Jidokai shouldn't be told only as a platform, but as a repeatable way of working for root-cause analysis, process stability and escalation.
This is where Lean Competence can make the decisive difference: senior practitioners bring experience in plant reality, change, standards and internal capability-building.
The takeaway
A pilot shows that AI can help. A rollout shows whether the organisation turns it into an operating system.