Two instruments, one of them lying
Where the agent's time goes
An agent that takes twelve minutes is not twelve minutes of work. Most of it is the model deciding what to do next, and the trace that is supposed to show you this can be computing the numbers rather than recording them. These posts open the instrument first, then use what survives to find out where the time actually sits — and why the intervention that helped did not make anything faster.
2 posts
Where this stands
- Held to be true
- Agent wall clock is dominated by model round trips, not by tools, and the useful target is the spread rather than the median. One intervention moved the spread from 8.8 minutes to 1.0 while barely touching the median, which is the shape worth chasing.
- What would break it
- A workload where tool execution, not the round trip around it, accounts for most of the elapsed time.
- Still open
- Whether a narrower spread is worth anything to the person waiting. It was measured because it was measurable, and nobody has asked whether predictability beats speed here.
- 1Evals5 min
The span that answered a question it never measured
OpenTelemetry LLM spans on every tool call told me the agent never ran anything in parallel, and I withdrew it. The numbers were not wrong. I had written them myself.
- 2Agents6 min
The median barely moved and the spread collapsed
Claude Code speed on code review: one line naming the search command. The median barely moved, the worst case fell from 18.0 to 12.2 minutes, and tool calls did not change.