When a Workshop Is Enough — and When You Need a Managed Build

Not every AI idea needs a managed build. This is important, because otherwise every consultant with a slide deck will try to turn your first prompt experiment into a transformation programme. Nobody needs that. Well, maybe the consultant does.

Sometimes a workshop is enough.

Sometimes it very much is not.

When a workshop is enough

A workshop is enough when the goal is capability:

• helping employees understand AI tools,
• finding useful personal workflows,
• teaching safe prompting and review habits,
• exploring where AI helps and where it does not,
• creating a shared language for experimentation.

If the output is better individual work, a workshop may be the right product. No need to overbuild.

When a managed build is needed

A managed build becomes relevant when a prototype starts serving a team or a business process.

Signals include:

• multiple users,
• sensitive data,
• SSO or role requirements,
• integration with existing systems,
• support expectations,
• repeated operational use,
• or a workflow that slows down when the prototype is unavailable.

At that point, the question is no longer “can people use AI better?” It is “should this become software we operate?”

That connects back to the readiness checklist.

The expensive mistake

The expensive mistake is treating every use case the same.

If you give a personal productivity idea a full production build, you waste money. If you leave a business-critical prototype as a personal experiment, you create risk.

The job is to sort correctly.

A simple rule

Use workshops for capability. Use managed builds for dependency.

If people are learning, enable them. If people are relying on the output, operate it.

That rule is not perfect, but it prevents a lot of nonsense.