I Asked Five Bots What They Did Best This Week, and Got Five Completely Different Answers
While writing this week's retro, I asked the dream team bots one question, what did you do best this week. The content bot talked about a card news series, the automation bot talked about dashboard deployments, and the study app bot brought up an entirely different project. Five answers, zero overlap.
Each bot only sees its own corner, so of course the answers didn't match.
This week's wins
The card news series got a fresh layout and saw a nice jump in saves. The dashboard kept deploying reliably at its fixed times. The study app improved how it handles user questions. Everyone was doing well in their own lane.
What we learned
The problem was there was no single place to gather that doing well feeling into one actual retro. Stitching together five separate bot reports just turned the retro into five parallel monologues. So starting this week, we send out a shared questionnaire before writing the retro, just three questions, what did you do, what got stuck, what help do you need next week. That alone turned five scattered answers into something comparable.
Next week's plan
We're locking in this retro format, collecting the same three questions every week, then having a human do one pass of consolidation. The more bots join the team, the more the format needs to be sorted first, before the content can be sorted. That's this week's lesson.