Build fast, trust slow
The fastest part of my week is the building. That is the part everyone wants to talk about - how much one person ships now, how the estimate that used to read sixty hours reads eight. It is real, and anyone calling it hype has not done the work. It is also the part I worry about least.
The part that decides whether any of it was worth doing is slower, and it has a name every manager already knows. Onboarding.
When a good person starts, you do not hand them the keys on the first morning and walk off. You give them real work, lots of it, earlier than feels comfortable - because that is how anyone learns a business. And then you read behind them for a while. Not because you think they are slow. Because they have not earned the context yet, and context is the one thing that only time buys. The reading-behind tapers as the trust builds. One day you notice you have stopped checking, and that day was earned, not granted.
That is exactly how I run an agent.
A lot of rope, almost no trust, and I only move the second slider as the first one proves out. Build fast, trust slow. The build is cheap now, so the saved time goes to the only thing that was ever scarce: verification. A check that grounds every figure against its source. A live run I watch in the room before it ever runs unwatched. A week of its output passing my own eyes before it passes anyone else’s.
None of this is caution for its own sake. Slow trust is how you earn the right to go fast later. The reps are not a tax on the speed - they are what makes the speed safe to keep. Skip them and the first silent failure does the teaching instead, and it picks the worst possible moment to do it.
A new hire and a new agent are the same management problem in different clothes. Teach generously. Verify relentlessly. Pull the verification back only when the work says you can, not when the calendar does.
Speed you can buy in an afternoon now. Trust still costs what it always cost - time, and the reps to back it.
Build fast. Trust slow.