Microsoft's AutoTailor compresses agent trajectories into 33 APIs, hitting 90.6% on WebArena
dair_ai · x · 2026-09-22
Microsoft Research's AutoTailor paper auto-generates MCP browser-automation tools from agent trajectories. Web trajectories become parameterized APIs, then a quality filter (wrong granularity/duplicates) and a usage filter cut 1,283 candidates to 87; online monitoring adds APIs for recurring gaps and prunes unused ones, converging at 33. On 106 WebArena Postmill tasks, 33 APIs plus a ReAct fallback reach 90.6% correctness vs 87.5% for ReAct alone, with request tokens down 57.8% and latency down 29.4%. Without the fallback, accuracy matches the unfiltered set using 94.9% fewer request tokens.
More from coding & agent
- Qwen3.8-27B spent 21 days building its own CUDA engine on one RTX 3090 — DanGrover · 2026-09-22
- Ontology layer for agents boosts GPT-5.5 by 26.7 points on DDR-Bench — omarsar0 · 2026-09-22
- Dev wires typesafe's Jev into Apple's Foundation Models framework for Swift apps — rxwei · 2026-09-22
- LangSmith ships Jev-as-a-judge to score every production trace cheaply — airesearch12 · 2026-09-22
- Garry Tan says Capy handles large PRs faster than Codex or Claude Code — garrytan · 2026-09-22
- Building an image rating tool with GPT Vision and Jev: what worked and what didn't — huangyun_122 · 2026-09-22