The series

Telemetry on Demand

What if telemetry was decided at runtime instead of at deploy time?

We record everything, all the time, because changing what we record has always meant a redeploy. These posts follow a different idea: lean telemetry by default, with OpenFeature flags turning detail up where something goes wrong, and agents fast enough to do the turning.

It’s a working theory, built in public a post at a time, and I’d like to hear where it breaks.