Training and enablement
Four rooms, four different conversations
An organisation does not adopt AI in one place. The board has to approve it, the managers have to find the work worth doing, the engineers have to build and run it, and everybody else has to use it without creating a problem. Those are four different sessions, and treating them as one is why most enablement spend does nothing.
Request a briefingExecutive and board
AI for the people who sign for it
Half day
What these systems can and cannot do, what they cost to run once the pilot budget is gone, and where the liability sits. Written for people who have to approve the spend and answer for the outcome, not operate the tool.
- What is genuinely new since 2023, and what is a rebadged capability you already own
- Where the recurring cost actually lands: inference, evaluation, integration, and the people
- How to read a vendor proposal, and the five questions that separate a real one from a demo
- What the EU AI Act and UK GDPR oblige you to, in the language your board minutes use
Managers and functional leads
Finding the work that is worth automating
One day
Most organisations do not fail at AI because the models are weak. They fail because nobody could name a process worth the effort. This is the workshop that produces that list, from the people who actually run the processes.
- A repeatable way to score a process on volume, variance, cost of error and cost of delay
- What to hand a model, what to hand a rule, and what to leave with a person
- How to write a brief an engineering team can act on without a second discovery round
- Measuring the thing afterwards, so the next one is argued from evidence
Engineers and technical teams
Production LLM and agent systems
Multi-week programme
The full engineering curriculum, hands-on against your stack. This is the track with real inventory behind it: the same material is taught as a cohort course, so you are buying something that has been delivered and revised, not drafted for you.
- Retrieval that survives real corpora, and the evaluation that tells you when it stops working
- Tool use, agent loops, and the failure modes that only appear under real traffic
- Observability, cost control and regression testing for non-deterministic systems
- Labs run against your own codebase, so the output is your team's code, not an example repo
All staff
Safe and useful daily use
Half day, repeatable
The session that turns a licence spend into a capability. Delivered alongside a rollout, so the policy people are asked to follow arrives with the reasoning behind it rather than as an email from legal.
- Getting a useful answer, and recognising a confident wrong one
- What may not be pasted into a general-purpose assistant, and why that line is drawn where it is
- The organisation's own policy, taught as part of the session rather than circulated separately
- Role-specific patterns worth keeping, gathered from the teams in the room
Tracks for a specific function are built to the audience when an engagement calls for it, working from your own processes. We do not list courses here that nobody has taught yet.
The shortest way in is half a day
It is the cheapest way for both sides to find out whether there is a longer piece of work here, and you keep the shortlist either way.

