I Built a CEO an AI Chief of Staff From a Barstool
By the time I left the bar, a CEO I work with had a working AI Chief of Staff — triaging his inbox, reading his calendar, remembering things about him. I never opened a laptop. Here's what that actually means.
Tonight I was at a bar. By the time I left, a CEO I work with had a brand-new AI Chief of Staff running on a machine in his office — triaging his inbox, reading his calendar, remembering things about him, and answering his texts. I never opened a laptop.
That sentence is either marketing fluff or the most important shift in how work gets done this decade. I think it’s the second one, so here’s exactly what happened and you can decide for yourself.
The founder’s real problem
The client is the CEO of a growing consumer brand. No IT department — he is the IT department, and he’s the first to tell you he isn’t technical. His problem is every founder’s problem: too many open loops, an inbox that owns him instead of the other way around, and no one whose entire job is making sure nothing important falls through the cracks.
A bigger team is the usual answer. It’s also expensive, slow, and adds its own management overhead. So we tried a different one.
Not a chatbot — a chief of staff
The thing we built for him isn’t a novelty that answers trivia. It reads his email and sorts it into what actually needs you versus FYI versus noise — with a one-line reason on every call, and it never moves or deletes a thing. It knows his calendar. Every morning at 7 it texts him his day, laid out. And he interacts with it the way he’d talk to a person: by text message. No app to download, no dashboard to learn, no account to create.
That last part matters more than it sounds. The bar for using it is the same as the bar for texting a friend, which means he actually reaches for it.
The part that makes it feel alive
The piece we added tonight is the one that turns a tool into a teammate: memory. Now when he tells it something — who his key people are, how he likes things handled, what he’s chasing this quarter — it remembers, and every answer after that gets sharper. Most “AI assistants” are goldfish; they forget you the second the conversation ends. This one doesn’t. That difference is the whole game.
I didn’t build it by typing code
Here’s the part I want you to sit with. I didn’t wire any of that up tonight by hand-writing code at a terminal. I built it by telling my own AI assistant what I wanted — from my phone, between drinks — and it did the work: the integrations, the security, the deployment. When something got blocked, it found another way in. I wasn’t doing the labor. I was directing it.
That’s not a party trick. That’s the new shape of the work.
The deal that just changed
For most of my twenty-six years in this field, the deal was simple: if you wanted the outcome, you needed the specialist who could personally do the thing. Wanted a network automated? Hire the network engineer. Wanted software? Hire the developer.
That deal is quietly ending. The people who win the next decade won’t be the ones who can do everything themselves — they’ll be the ones who know exactly what they want and can direct capable systems to build it. Taste, judgment, and knowing what “good” looks like go up in value. The grunt work trends toward zero. The bottleneck stops being can someone build this and becomes does anyone know precisely what should be built.
If this is you
If you’re a founder drowning in your own operation, you probably don’t need a bigger team. You need one sharp system that actually knows your world and takes the load off — built around you, not around a generic template.
That’s what I do now. And apparently I can do it from a bar.