Built in a real firm, not a lab

I’m Craig Hobbs. I spent years in banking operations before starting a bookkeeping firm built for real estate investors. The firm grew from two clients to dozens of entities and 77 sets of books — and somewhere in that growth curve I hit the wall every firm owner hits: the work scales linearly with headcount, until it doesn’t scale at all.

So I built a fleet. Not a chatbot, not a demo — an organized crew of AI agents with names, jobs, standing orders, and a chain of command. They pull bank statements, rebuild transaction histories, clear review queues, prep reconciliations to the penny, and file their own work reports. The humans in my firm do judgment; the fleet does typing. It runs while I sleep, and every morning I read its overnight report like a general reading dispatches.

The command agent at the center of mine is named Alfred — because every Batman needs an Alfred, and I’ve yet to meet another firm that has one. When I demoed the fleet to a room of fellow bookkeeping-firm owners, the questions didn’t stop for an hour, and one of them was running his own first bot on real client files within days. That room is why Pennyworth exists.

What Pennyworth is

The teaching side of the operation: a community where firm owners learn to run their own fleets, and a white-glove service for the ones who’d rather have it built for them. Everything taught here runs in production in my own firm first — on real client books, verified against real bank statements, with receipts for every claim.

What it isn’t

Hype. I don’t sell “AI will replace accountants” and I don’t sell magic. Bots do mechanical work under doctrine; humans hold the judgment and the client relationships. That division of labor is the entire method — and it’s why it works.

Hobbs AI Consulting, LLC