Tools get bought, pilots get launched, and a year later the work still moves the same way and nobody can say what changed. Six stages, gates you cannot talk your way past, and software that keeps the score.
Whoever is answerable for how AI is going across the business: often a chief operating officer, a transformation lead or a chief people officer.
Consulting with accountability between the sessions. Somebody is answerable for each pilot by name.
It runs in TransformLayer, so every opportunity, decision, owner and result is somewhere you can both open.
Giving everybody a licence distributes access. It does not build capacity. Usage becomes the only number anybody reports, and nothing was baselined before it started, so there is no honest way to say afterwards whether it worked.
So measurement comes before pilots, gates are enforced rather than suggested, and the ledger keeps measured apart from claimed.
People Team AI trains the people. The PROVEN Framework™ installs the method. TransformLayer runs the programme.
Practical AI training for People teams, up to an Applied AI Champion Programme that ends in evidence rather than attendance.
Explore trainingConsulting that sets up how your company finds, filters, pilots and measures AI ideas, ending in scale, change or stop.
AI use case management software. One place for ideas, approvals, pilots and what each one returned.
Explore TransformLayerUse one, or all three. Training works on its own. TransformLayer is bought on its own. The PROVEN Framework™ is there when you want help setting the whole programme up properly.
One story the whole board tells the same way
A signed charter and a two-sided measure: one commercial, one people. If the executive team cannot tell the same story, nothing downstream survives a budget round.
Gate 0: the charter is signed before anything else starts.
What is already happening, and what can actually be measured
A census of the AI people already use without telling you, data and people readiness, risk boundaries, and the one most programmes skip: can this actually be baselined?
A scored register, not a wish list
A task-level census across twelve functions, scored into a register you can defend. Not sticky notes, and not a vendor telling you what their product happens to do.
Gate 1: no baseline, no pilot. Anything without one gets a task to go and get one.
Ninety-day pilots owned by line managers
Owned by the people who run the work, not a central team. Named champions, signed guardrails, and usage somebody actually looks at.
Gate 2: scale it, change it or stop it. Stopping it is a real outcome.
Three numbers, kept apart on purpose
Hours released, cash converted and claimed but not realised, never added together. Every figure tagged measured or estimated, and the sponsor signs the pack.
Reinforce, retire, scale, rotate, repeat
Quarterly review, champions rotated so it does not depend on two enthusiasts, and back to Readiness. An AI programme is not a project with an end date.
Every methodology has stages. What is unusual is that you cannot walk past a gate because the quarter is ending. No signed charter, no programme. No baseline, no pilot. No honest evidence pack, no scaling.
That is uncomfortable, and it is the entire value. A gate you can talk your way through is a suggestion, and suggestions are what produce a year of activity with nothing to show a board.
Most consulting leaves you with documents. By month four the spreadsheet has three versions and the benefits case is a slide somebody wrote in March.
Here, hours released, cash converted and claimed but not realised stay three separate figures, every result is labelled measured or estimated, and finance signs off the ones that are real. Most programmes report the third as though it were the second, which is how a business believes in a return nobody can find.
A programme that runs on outside people ends when those people leave. This one is built to be handed over from the start.
The Applied AI Champion Programme develops the internal capability this method needs, and the applied assessment is the evidence those people can use it on real work.
The method gives them a job: every suggestion is triaged by somebody who knows the work, and the ones that survive get scoped properly.
Most ideas should not be built. Some are a rule, not AI. Some save an hour a year. Some are a decision about a named person and should never go near a model. A champion who can say no with a reason is worth more than a backlog of enthusiastic pilots.
This is also why champions do not tread on a CTO, a security lead or a data team. They are a buffer, not a rival.
Fewer conversations with engineering, data and security, and better ones. The expensive part is temporary by design: I install the method and train the people who run it.
The method is written and the stages are fixed. What does not exist yet is a shelf of case studies, and this page is not going to pretend otherwise.
Early engagements are shaped with the client: the method applied to your organisation rather than a template applied to you, and a price that reflects it.
Tell me roughly how big the organisation is, who is sponsoring the AI work, and what you are being asked to show for it. If a workshop or a quarter of tidying up your own measurement would serve you better, you will be told that, because a programme sold to somebody who was not ready is the one that fails publicly.
Email me directly , or use the enquiry form and say it is about the PROVEN Framework™.