I get sales and ops teams actually using AI, and build the tools they need when nothing off-the-shelf does the job.

Fair question, and most people selling this can’t answer it. The AI runs through an account your business owns, ringfenced from everything else. Every conversation is logged, so you can see what’s being asked. Nothing is trained on your data, and nothing goes anywhere you haven’t signed off. It’s written down before we start.

Most companies this size haven’t. The tools live in accounts you already own, and the person who owns the job owns the tool: the ops manager runs the Monday report, not a developer. Anything with code in it is written up for whoever does your IT, in their language. If you’d rather I stayed on the end of Slack, that’s the ongoing option, and it’s for the next thing, not for keeping the last one alive.

That’s the usual shape of it, and it’s normally the owner or the one technical person. You’ve done the hard part already: working out what it’s good for. What’s missing is everyone else, and the things that only exist if someone builds them. You’re the top row of that diagram up the page. The job is moving the middle row up to you.

Nearly everyone does. Meeting summaries nobody reads, an autocomplete nobody switches on. Paying for it isn’t using it.

Most training stops at basic prompting, so six weeks later everyone’s using it as a search box. A workshop is a day out with generic examples, and everyone’s back to normal by Thursday. Rollouts usually die with middle managers: execs approve it, managers quietly make it optional. Underneath all three there’s often a quieter worry too: if AI does the work, what am I for? So I don’t run a session. I work with the team on their own work, the people who’d otherwise opt out choose what we fix, and I stay long enough to watch the new way wobble and fix it while I’m still there.

They would. That’s what happens when someone senior comes back from a conference converted and tells everyone to use AI. I start the other way round: your team says what they’d rather not be doing, and we fix that first. Nothing gets added to anyone’s plate in week one. And nobody has to admit they don’t get it. Most people are making this up as they go along, and taking the hierarchy out of the room is half the job.

Eventually, do. But a good one is months away and expensive to get wrong, and they’ll land in workflows nobody has fixed yet. If you’re hiring anyway, this is the best time: new starters learn the new way from day one, and the answers they’d otherwise have to ask you are written down where they can ask them.

I’ve held that seat, so I know the difference. A fractional CTO owns the whole estate: the roadmap, the suppliers, the systems, the IT when it comes to it. This is one half of that job. I sit with sales and ops, find the work eating their week, build the tools that do it and hand them over. Whoever runs your technology keeps everything else, and if what you actually need is the whole estate run, I’ll say so.

Good. They’re probably answering every AI question in the building already, on top of the job, and lately having their answers checked against ChatGPT. I’ve been that person: getting AI into the engineers was the easy half, and I never had the weeks to sit with sales, ops or marketing. I’m not a replacement. I’m the three months they don’t have, and what I build is theirs to run afterwards. More for technical leads →

It will, if a chatbot drafts the email and someone hits send. What I build starts from your team’s own words: last year’s quotes, the handover you’d write if you had the time, the report in the shape the board already reads. The AI does the assembling. The person does the judging. If a customer can tell, we’re not finished.

Then I build it. That’s the half most people offering this can’t do, and it’s usually where the actual gain is.