A practitioner's rule for scaling agent autonomy: start at one issue a day

Pocock posted a short rule for introducing an agent into a triage workflow: "When you first start building a software factory, make it TINY. Don't get it to triage every issue in your repo. Just do one per day. Check its output. Adjust it. Keep going with your regular work. As you begin to trust its judgment, get it to triage TWO issues per day instead. Then five. Then ten. Start small. Adjust. Build trust. Expand." He credited the idea to Dex Horthy, founder of HumanLayer, who publicly documented his own factory's growth outpacing its guardrails in an earlier talk. Posted at 7:02 a.m. Eastern, the post had reached 28,800 views, 650 likes, and 317 bookmarks within roughly two hours, a pace of about 300 likes an hour. Six minutes after it went up, Horthy quote-posted it himself: "this guy gets it. This is as much a @0xblacklight innovation as a dex one," crediting @0xblacklight as a co-originator.

Why it matters: Pocock's one-per-day ramp treats human review capacity as a constraint on agent autonomy. It gives a reviewer time to catch a bad pattern before the agent receives more of the queue. Horthy has drawn the same lesson from watching his own setup grow beyond consistent human review.