AI maturity check
Seven questions about how your firm actually uses AI. Three minutes, and the report is free.
Compare yourself to other firms in the investment industry. It is a conversation, not a form — answer in your own words.
The AI Maturity workshop · Two days
You have 200 licenses.Forty people use them.
Nobody lied to you. The rollout genuinely happened — the seats are provisioned, the training was delivered, the invoice arrives every month. What didn’t happen is the part nobody scoped: the work itself changing.
Sound familiar
Which one is your Monday?
“We’re paying for 200 seats. Finance asked me last week what we got for it.”
You can’t answer because nobody instrumented it. Usage data tells you who opened the app. It doesn’t tell you which process got faster, and that’s the only number Finance recognises.
“We trained everyone in January. By March it had fallen off a cliff.”
Training creates capability. It doesn’t create obligation. Six weeks later people revert to the workflow their deliverable is actually built around — because that workflow never changed.
“We moved off ChatGPT in the spring. Same people, new logo, same three prompts.”
Switching vendors is a procurement event. Your team carried their old habits across intact, and the new tool’s real capabilities — projects, skills, agents — went untouched.
“I can see exactly what this does. I’m one person, and I have a day job.”
You’re the firm’s entire AI capability, running as a side project. That’s not a strategy, it’s a dependency. The fix isn’t more of your evenings — it’s ten more people who think the way you do.
“The proof of concept was great. It’s been a proof of concept for nine months.”
Pilots stall where responsibility gets vague — the moment the thing has to survive a real deadline, a compliance question and someone’s PTO. That’s a design problem, not an enthusiasm problem.
“Everyone says they’re saving time. I can’t find it anywhere.”
Time saved in fifteen-minute slices, spread across forty people, disappears. It only shows up if you consolidate it into one process, measure it end to end, and then reassign what you freed.
“Two of our competitors are talking about agents on panels. We’re still evaluating.”
Evaluation is comfortable, and it’s the most expensive thing on this list. The gap compounds — not because they have better models, but because their people have twelve more months of practice.
The question
“They just need training on the tools.”
The problem is the question, not the tool.
Nobody notices, because it is the question everybody types.
Four words, no task in them, no reader, and no standard for what a good answer would look like. What comes back is thin, and the tool takes the blame. Maturity is not knowing more about the model. It is turning the question you actually have into one that can be answered, then knowing how hard to check what comes back.
What it looks like
Four habits, and none of them are prompt tricks.
These are what we watch for in a room. They arrive together or not at all, so there is nothing here to work through in order.
Leaving it alone
The first thing that changes is what people stop handing over. The biggest single gain in most rooms is somebody learning to recognise the half-hour a model will confidently waste for them.
Saying the whole task
Most bad output is a badly stated request — no reader, no format, nothing that says what finished means. That is technique, it is teachable, and it is the part that gets skipped.
Checking in proportion
Not re-reading everything in a panic, and not trusting an answer because it reads well. A checking habit sized to what it costs you if it turns out to be wrong.
Working where people can see
Usage hidden from a manager never compounds. Teams mature when a failed attempt is something you show the room rather than something you quietly delete.
How we teach it
Two days, on the files you already have open.
No case studies and no fictional companies. The people who do the job bring work that is already on their list, which means half the room finds out on the first morning that their best example is one a model should not be touching.
The other half of the workshop has nothing to do with tooling: what your organisation currently rewards, and whether anybody can change how they work without being punished for it. Skip that half and the maturity leaves with the people who were already curious.
See the two days against everything else we doTell us where your team keeps stalling.
What they are already trying to do with AI, and the point where it stops being worth the effort. If two days is the wrong shape for it, we will say so and tell you what is.