AI team enablement and training
The goal is a team that uses AI well without me in the room. That means teaching which work to hand over, which work to keep, and how to tell when the output is quietly wrong.
What it is
Training that ends with your people doing the work. Most AI training is a demo, and demos do not survive contact with a Tuesday. What changes behavior is sitting down with the jobs your team actually does, deciding together which of them AI should touch, writing that decision down, and then watching people do the work until it sticks.
Three things get written. A list of the work the AI drafts. A list of the work it never touches, usually anything involving a price, a promise, or a person's private information. And the check that has to happen before anything leaves the building. The third one is the part every rollout skips, and it is the reason most of them quietly stop after a month.
Generation is cheap. Verification is the product. That is the whole lesson, and it took me a hundred and twenty-five published pages to learn it properly.
What I've actually done
I learned this the expensive way, by shipping. Tiller's content pipeline publishes AI-written pages, and the useful lesson was not about prompting. It was that every page needed a gate: 125 guide pages are live, and each one was checked against an expert-ruled source matrix and a voice test that rejects off-brand output before a reader ever sees it.
On a client engagement this year, the rule that mattered most was about approval, not about AI. Nothing edits the live store. Drafted copy goes into a sheet for row-by-row approval by the owner, because it is his business and his voice. The AI made the work possible at that volume. The approval step is what made it safe to use.
I also measure rather than assert. In August 2026 I ran a baseline study on 18 established firms in my own county, asking an AI assistant with live web search the 30 questions their customers ask. Nine of the 18 were never named once. That is the kind of before-and-after number I would rather hand a team than a productivity claim.
How an engagement runs
- A working session on the real work. Bring the jobs, not the tools. We sort them into draft, never, and check.
- Rules on one page. Short enough that people read it, specific enough that a new hire can follow it.
- A supervised run, then a check-in. Your team does the work with me watching, and we look again two weeks later at what actually happened.
Who it's for
Small teams where one or two people are already using AI on their own and the rest are unsure or quietly against it. Also owners who bought a subscription for everybody and cannot tell whether anything changed. Both are normal, and both are fixed by the same thing: written rules and a check, rather than more enthusiasm.
Questions I get asked
How long does training take?
A working session and a follow-up, usually across two to three weeks. It is short on purpose, because the rules only stick once people have used them on real work.
Which AI tools do you teach?
Whatever you already pay for. The habits transfer between assistants, and the rules we write are about the work rather than about one vendor's interface.
What if my team is resistant?
That is often the healthy reaction. Naming the work AI will never touch, out loud and in writing, does more for adoption than any demo.
Do you need to be kept on afterwards?
No. Handoff is the deliverable. You keep the rules, the documentation, and a team that can run without me.