NavigateMENU / 04
More
Elsewhere
AI Consulting
For teams standardising on Claude, or already there and going deeper.
The gap inside your company is wider than the gap between companies.
One person on your team has built four internal tools and hits their usage limit every Thursday. Someone two desks over uses it to check the commas in their emails. Neither knows what the other is doing. Nobody has told leadership, and leadership is still pointing at the chatbot on the website.
Premise
Nobody in your organisation knows how wide the spread actually is, including the people sitting at the top of it.
We started running these sessions because clients kept asking us to build them an AI tool, and the tool was almost never the problem. Four people would have used it well. Six had no idea it existed. Two of them had already built something similar in a spreadsheet months earlier, and hadn't thought to mention it to anyone.
That's a measurement problem before it's a software problem. Find out where everyone actually sits, then do the work to move the floor.
The Spread
Four positions on one team.
Every team we've measured splits roughly like this. The labels matter less than the distance between the ends of the list, and the fact that almost nobody in the building can see it.
01
Hasn't started
Has an account, opened it twice
Usually not stubbornness. They've seen demos of it writing poetry and haven't been shown a single thing that touches their actual job. Nobody has ever sat with them for twenty minutes.
02
Uses it as a better search box
Asks, reads, closes the tab
Genuinely useful, and roughly a twentieth of what the licence costs. This is the largest group in most organisations and the one with the most room to move.
03
Uses it for real work
Drafts, rewrites, first passes, analysis
Has opinions about which model to use for what. Would notice immediately if you took it away. Rarely writes any of it down, so none of it spreads.
04
Builds with it
Projects, Skills, MCP, small internal tools
Hits the usage limit most weeks. Has quietly stopped explaining what they're working on, because the explanation takes longer than the work. This person is the most valuable and the most isolated employee you have.
Meanwhile, upstairs
The chatbot went live and everyone moved on.
We build those. When the retrieval is solid and the scope is honest, they earn their keep, and we've written a whole field guide on doing them properly.
What a website chatbot doesn't do is change how one person on your team works on a Tuesday. An AI strategy that lives entirely on your public site hasn't touched your operations at all. Meanwhile the licences are being paid for every seat in the building, and most of those seats are using a fraction of what they cost.
If the thing you actually need is systems work, integrations, agents and tooling, that lives here: the AI integration field guide.
The Engagement
How we run it.
Four to six weeks, start to finish. Two workshops, one leadership session, and a report that shows you a number that moved rather than a page of opinions about how it went.
- 00Week 1
Baseline
Everyone on the team fills out a scored questionnaire. Where they sit, what they use it for, what they've tried and abandoned, whether words like Projects, Skills or MCP mean anything to them. It takes about ten minutes and it's the only way to find out how wide your spread really is.
You getA scored distribution of your team
- 01Week 1 to 2
Build
We take what the baseline showed us and build two to four working tools against tasks your team already does. The reporting job someone loses every Friday afternoon to, built so they can watch it run instead of imagining what it might look like.
You get2 to 4 working tools, yours to keep
- 02Week 2
Workshop I
We split the room by fluency so nobody sits through basics they mastered a year ago, and nobody gets buried under vocabulary they've never heard. Then we demonstrate the tools we built for you and hand them over. Anyone who wants to attend the other group's session is welcome to.
You getStreamed sessions, tools handed over
- 03Week 2
Leadership session
Forty-five minutes with whoever funds this. Not a fluency session. What it costs, what it won't replace, what a reasonable expectation looks like six months out, and why the chatbot wasn't the finish line. This is usually the hour that decides whether any of the rest of it sticks.
You getExpectations set with the people who set budgets
- 04Week 2 to 4
Practice
Everyone leaves with assigned work, pitched at the group they were in. It lands somewhere shared rather than in a private chat window, so people can see what their colleagues are building. This is the part that fails everywhere else, which is why we give it a deadline and an audience.
You getAssigned work, posted where the team can see it
- 05Week 4
Workshop II
Opens with two or three of your own people demonstrating what they built. Not us. Watching a colleague solve a problem you also have does more for adoption than anything we can say standing at the front of a room. Then we extend the tools and go a layer deeper.
You getPeer demos, extended tooling, deeper capability
- 06Week 5
Re-measure
The same scored questionnaire, run again. Then a written report: what moved, what didn't, what your team said in their own words, who should own this internally, and what to do as the tools underneath you keep changing.
You getA before-and-after distribution, and a written report
- 07Week 10
Check-in
Thirty minutes, about a month after the report lands. Relapse happens here, once the novelty has worn off and everyone's calendar has filled back up. We look at what's held, what's slipped, and whether the person who owns it internally has what they need.
You getA held floor, or an honest account of why it slipped
What you're left with
The point is that it survives us leaving.
Illustrative shape, not a client result. Your numbers are whatever your numbers turn out to be, which is rather the point of measuring them.
The tools, and the ability to change them
The two to four tools we build are yours outright. Part of Workshop II is your people modifying them in front of us, because a tool nobody can edit becomes dead weight the first time your process changes.
A named internal owner
We pick this person with you during Workshop I. Every rollout that survives has one, and if you don't name them, nobody is. Part of the final report is written to them specifically.
A distribution that moved
The same instrument, run twice, so you're holding evidence rather than impressions. This is also the document your team lead can put in front of the board without translating anything.
The instrument itself
You keep the questionnaire. Run it again in six months, after you've hired four people, and you'll know within a day whether the floor held.
Who this isn't for
We'd rather say it now than on week three.
- Teams not consolidating on Claude. We work deepest in that ecosystem and the tools we build assume it. The habits transfer to other models, the specific tooling doesn't.
- Teams under about eight people. Book the call anyway and we'll tell you what to do in twenty minutes for nothing. It genuinely doesn't need a sprint.
- Anyone who wants a keynote. There's no stage, there's homework, and the work only lands because people leave with things to do.
- Anyone whose real problem is that the software doesn't exist yet. That's a build, and it's a different conversation.
The productized version
The AI Fluency Sprint
Everything above, one fixed price.
The full engagement, four to six weeks, scoped and priced so you can forward it to whoever needs to approve it without booking a meeting first.
From $4,950
See what's included →After the sprint
Moving the floor is one thing. Holding it is another.
Six months on, three people have left, two have joined, and the models underneath you have shipped capabilities that didn't exist when we ran your workshops. What holds a floor in place is infrastructure your team owns, with the data consolidated somewhere they can all reach it. That's a business operating system, and it's the engagement that tends to follow this one.
Business operating systems →Start here
Tell us where your team actually is.
Twenty minutes. Bring the situation rather than a brief. If you're the person trying to get your organisation to take this seriously and it isn't landing, say that out loud on the call. It's the most common reason people book one.
Book a call →