We write about the parts that were hard

Conveo runs AI-moderated qualitative research: real conversations with real people, at a scale that used to require a room full of moderators. That means a voice stack measured in milliseconds, an analytics layer that has to explain itself, and a lot of decisions that only look obvious in hindsight.

This is where we write those decisions down — the architecture we picked and why, the thing that broke in production at 3am, the measurement that changed our minds. No launch posts, no thought leadership. Just engineering notes with enough detail to be useful to somebody solving the same problem.

What you will find here

• Post-mortems — what broke, what invariant we had wrong, and what we changed so it stays fixed.
• Architecture notes — the tradeoffs behind the pipelines, and the options we rejected.
• Measurements — numbers from production, not benchmarks from a laptop.

Following along

Everything here is in the RSS feed. If you want to talk about a post, the fastest route is LinkedIn or the team page at conveo.ai.