By the time Kirsten and I were out on the deck shala this morning, with the keto dirty chai, my company had done three things and I had done none of them.
A reconciliation routine had read overnight bank data for an enterprise client and drafted the month's entries, setting aside the handful it wasn't confident about. One of my own products had read its own traffic from the day before and written a small change to itself. And a client call from the previous afternoon had turned itself into a filed report, branded, in my voice, sitting in the right folder waiting for me to read it.
I noticed this only because I had sat down to write something about it. I'd spent the morning drafting a long pillar page trying to define, properly, what an AI-first company is and how you become one. I was halfway through a section on what the operating day looks like when I realised I was describing my own Monday and hadn't clocked it.
That's the strange thing about how to run your company with AI. You don't feel the moment it happens. There's no launch. One week you're doing the reconciliation by hand, the next week a system does the draft and you review the exceptions, and a few months later you genuinely could not go back to the old way because there is no team standing behind the system waiting to pick up the work if it stopped.
The pillar I was writing draws a line between two terms people use as if they're the same. AI-first is a stance: you decide AI is the default starting point for getting work done. AI-native is a structure: the company is built so that if you removed the AI, it would stop working. Most companies can only go AI-first, because they already exist with their staff and their habits. You become native by accident, one workflow at a time, until the old version isn't possible any more.
Writing it down made me see I'm closer to the native end than I'd have said out loud. Take the AI out of the reconciliation, the content pipeline, the meeting reports, the daily prospecting, or the growth loops on my own products, and each one simply halts. There's no fallback team. That's not a boast. For a one-person company it's just the maths. The work that would have justified my first three operational hires gets done by systems, so I stay profitable at a size that should be losing money on overhead.
I think this is the part that matters for anyone reading who did well in a career, saved hard, and is now staring at the idea of building something small on their own terms. The old fear was that you needed a team before you could build anything real, and a team meant payroll, and payroll meant you were back in the thing you left. That isn't true any more. The thing that used to require five people now requires one person and a stack of systems they trust. The catch is that you have to actually build and trust the systems, one at a time, on real data, which is slower and more boring than the demos suggest.
What stays with me is what's left on my plate once the volume work is gone. Not less work. Different work. The exceptions the systems can't call. The client relationships. The decisions about what to build next and what to say no to. The judgement. I spend more of my day thinking now than I did when I was doing the operational grind by hand, and that's the whole point of the trade. The boring work goes to the system. The work that needs a person comes back to the person.
I put the full version up as a pillar page yesterday: what an AI-first company is, how to become one, and what it looks like from the inside. If you want the discipline that sits underneath it, the AI operations playbook is the day-to-day, and the fractional Chief AI Officer pillar is about who should own the direction. The same one-person, AI-first setup runs the nonprofit side too, including the work behind the Finn Wardman World Explorer Fund.
Three things before frothy masala chai. None of them mine. That's the company I wanted to build, and I only noticed I'd built it because I sat down to write the definition.
Monthly Revenues $9,200 | Clients 2 | Prospects (Meta to WhatsApp campaign built, WABA pending) | Team: Me + Jan (CTO)