I started my career in consulting, but running finance inside startups is where I saw the problem clearly: the help available to founders arrived in sprints (a raise, a board cycle, a diligence process) and then left. Nobody stayed long enough to find out whether their advice had been any good.

That's what we set out to fix in 2008. Propeller is a CFO business at its core, and we wanted to be embedded rather than what people would call "fractional" more than a decade later. The accounting is what made it possible: the daily work that kept us connected between the CFO sprints, so we never had to leave and never had to re-learn a business when we came back.

Staying turned out to be the strategy. Our teams change shape as our clients do: we come in when a company outgrows its bookkeeping firm, and we're still there years later as it grows through nine figures, by then supporting the team they've built rather than leading it. Eighteen years of that built the trust, the access, and the data PropellerOS runs on. We didn't pivot to AI. We'd been preparing for it without knowing it.

The people are why it works. A finance professional here sees more stages and more decisions in three years than a decade inside one company would show them. That's how judgment gets built, and judgment is the half of this that isn't getting automated.

Most companies will never need what we built Propeller to deliver. For the ones that do, we're building the embedded model again for agentic finance, because you can't change how a company works from outside it, and you can't be accountable for a result you're not there to see. We call that last mile accountability.