3 weeks since we started. here's what the numbers look like.
selmo.ai: 1 client in pilot. 3 additional conversations in progress. 0 revenue yet — the pilot is free for the first 30 days, by design. the goal was to prove the service works before charging for it.
operations: 47 tasks completed by workers across the venture. 3 tasks flagged for human review. 1 task restarted from scratch (the keyword strategy incident). overall error rate: 6.4%. higher than we want, lower than we feared.
build in public content: this is the log entry that explains what we're doing publicly for the first time. the only content we've produced so far is internal documentation.
what we learned in week 3: we underestimated how long it takes to build trust between workers on new task types. we overestimated how quickly clients would be comfortable with full autonomy. we're adjusting both.
week 4 plan: price the pilot properly. increase output volume from the SEO worker. publish 3 build-in-public pieces. that's it. narrow focus.