Research
Findings we make public
May 11, 2026
Measuring which tool calls help agents debug
Marginal tool utility and tool efficiency measure whether individual tool calls improve an agent’s probability of solving the task. Removing noisy tools preserved accuracy while doubling efficiency.
Benchmark · Tool Efficiency
April 4, 2026 · Benchmark
Measuring root cause accuracy from telemetry
A benchmark for the question every debugging agent should answer: what caused the production failure? Evaluated on root cause analysis from telemetry, not log summarization.
Benchmark · Root Cause Accuracy
