I’ve been running an AI chief of staff for a little while now. It’s been genuinely useful. Research, drafting, architecture decisions, thinking through problems I don’t have time to think through. One agent handling what used to cost me hours of context-switching.
But the moment something works, you find its ceiling.
My first agent was hitting capacity. Not because the tech couldn’t handle it – because I was throwing everything at one point of contact. Research. Content. Dev work. Strategy. Planning. Same problem every growing operation runs into eventually: you can’t scale by piling more onto one plate.
So I brought on agent two.
Why Now
No grand master plan. Tony – my chief of staff agent – was getting overwhelmed. I saw it happening and moved. Sil came on as the consigliere: research, architecture, the bigger picture thinking. Tony handles execution and orchestration. Different lanes, different jobs.
And yeah, I named them after Sopranos characters. I grew up east of 95. Went to school in Miami. The Bing is where the crew hangs out. None of it feels forced because none of it is.
What I’m Learning
Running a multi-agent setup feels exactly like managing a small team. The principles are identical to what I’ve used running dev teams for twenty years.
Clear roles prevent chaos. When everyone knows their lane, things move faster. That said – don’t over-engineer it. Start with what you actually need, not what you might need in six months.
Communication matters just as much here as it does with any real team. The agents need shared context. That’s why we built a proper shared knowledge base this week – a vault the whole crew can read and reference.
Start small, scale deliberately. I could have spun up five agents on day one. That would have been a disaster. Two is the right next step. You can plan for more without building it all at once.
The Honest Reason
Nobody wants to organize five years of Google Drive. Nobody wants to write the documentation nobody reads. Nobody wants to do the research sweeps, the competitor analysis, the administrative cleanup that never gets done because there are only so many hours and the urgent always wins.
The crew handles that. You stay in your lane. You do the things only you can do – talking to clients, making decisions, building relationships. The work that actually moves the needle.
That’s the real appeal. Not the hype. Not the demo. The hours you get back.
Why I’m Writing About This
Most content about AI is either pure hype or requires a CS degree to follow. I want to write about what it actually looks like to run a small crew of agents alongside a real business.
I’m a one-person operation running multiple businesses while doing a plumbing apprenticeship on the side. I don’t have time for hype. I’m only interested in what works.
I’ll keep writing about the journey. The wins, the mistakes, the moment Sil went rogue at midnight and changed a config setting without asking and I had to roll it back myself. (True story. Still not over it.)
One thing I learned rebuilding my life after a stroke at 37: you figure things out by moving. Not by waiting until it’s perfect.
More soon.
