Part of Change Management & UpskillingAvailable on its own
Licences give your people access. The Lab gives them capability.
The Lab is where capability actually gets built — on your problems, in an environment you control.
Most organizations now hold more AI licences than they have people who can use them well. Training usually means a course: a few hours of demonstrations, a certificate, and work that carries on exactly as before. The Lab is the opposite shape. Teams come with a process they own, build working AI workflows against it over the programme, and leave with those workflows running. What gets built belongs to you, and so does the capability that built it.
The Lab is how we deliver upskilling inside a wider engagement, and it can be bought on its own — a single team, a single process, no broader programme required.
The offline environment
Offline models, single tenant, your data stays inside it.
For organizations that aren’t ready to put their people on public model services — and shouldn’t have to wait.
Plenty of organizations have a good reason to hold back. The policy isn’t written, the data classification work isn’t finished, or the risk position simply hasn’t been settled yet. That’s a reasonable place to be, and it usually stalls capability-building for a year or more while the decision works its way through.
The Lab removes the dependency. It runs offline models in a single tenant on Google Cloud, provisioned for your organization alone. Prompts, documents and outputs stay inside that tenant. Your teams start building now, on infrastructure you control, and the licensing decision gets made on its own timeline rather than blocking everything behind it.
| Hosting | Single tenant on Google Cloud, provisioned per organization |
|---|---|
| Models | Offline, running inside the tenant |
| Your data | Prompts, files and outputs stay within the tenant |
| What you keep | The workflows your teams build, exported at the end |
| Alternative | Programmes can run on your existing enterprise AI licence instead |
For the leadership team
Most executives have tried AI once and concluded it’s a better search engine.
The shift is from asking it questions to handing it work.
The usual first encounter is a summary of a document, a passable draft, and a quiet conclusion that it’s useful but not important. That judgement is reasonable given the input — a question with no context attached will produce a generic answer every time.
The shift is learning to brief it the way you’d brief a chief of staff: the constraint, the audience, the decision it’s feeding, what good looks like and what has already been tried. Executives who work that way stop using AI to look things up and start handing it work that used to sit in their own week.
There is a second reason to spend the time. A leader who doesn’t use AI well can’t judge which proposals in front of them are credible, can’t set a direction their organization will believe, and can’t model the behaviour they’re asking for.
What it covers
- Briefing with context: what to hand over, and what to hold back
- Delegating a recurring piece of your own week, end to end
- Judging AI work: where it is reliable, where it needs a check, where it should not be used at all
- Reading an AI proposal from your own teams and knowing what to ask
A small working session with your leadership team, on real work each of them brings. Not a demonstration.
For the teams doing the work
People learn this by building something they need, not by watching someone else build.
Cohorts work on their own processes and keep what they make.
Each cohort arrives with a process they own and live with — the reconciliation that eats two days a month, the report nobody trusts, the approval chase, the same exception explained again. Over the programme they build working AI workflows against it, with our operators and engineers alongside them.
What comes out is not a certificate. It’s a small portfolio of workflows that run, documented well enough for the team to maintain and extend without us. Because the work is theirs and the problem was already costing them, adoption isn’t something that has to be driven afterwards — the team has been using the thing they built since the week they built it.
Workshops run either in the Lab’s offline environment or on the enterprise AI licence you already hold, whichever fits your current position.
What a cohort leaves with
- Working workflows against their own processes, running in production or ready to be
- The skills to build the next one without us
- A clear-eyed view of what AI should not be pointed at
Where this sits
The Lab is one of four disciplines.
Upskilling is the fourth. The other three — foundational architecture, process redesign and AI enablement — are what make the workflows your teams build in the Lab worth running at scale. Many clients start here and work backwards.
Start here
Bring one process and one team.
The simplest way in is a single cohort working on a single process. We’ll help you pick one where the manual effort is already measurable, so the value of what gets built is obvious to whoever has to approve the next one. If the licensing position isn’t settled yet, we’ll run it in the Lab’s offline environment and that decision can wait.