{"version":1,"type":"story","url":"https://digestai.news/story/dereact-improves-weaker-ai-agents-with-external-gating-policies","json":"https://digestai.news/story/dereact-improves-weaker-ai-agents-with-external-gating-policies.json","markdown":"https://digestai.news/story/dereact-improves-weaker-ai-agents-with-external-gating-policies.md","slug":"dereact-improves-weaker-ai-agents-with-external-gating-policies","headline":"DeReAct improves weaker AI agents with external gating policies","summary":"Researchers introduced DeReAct, a modular agent architecture that separates action validation and task completion control from the main LLM policy. The approach uses a Critic to validate actions before execution and a Context Manager to verify task completion. Testing on GAIA and SWE-bench Verified showed DeReAct improved Pass@1 scores by 6.5–7.0 points for Qwen3-Coder-480B and 4.2–5.2 points for Claude Sonnet 4.5, with smaller gains for stronger models. With Claude Opus 4.5, performance remained similar to ReAct but produced more evidence-complete trajectories. The authors note external gating is effective when targeted failures are common and the gating policy is reliable.","keyPoints":["DeReAct uses a Critic to validate actions and a Context Manager to certify task completion","Pass@1 improved by 6.5–7.0 points for Qwen3-Coder-480B on GAIA and SWE-bench Verified","Gains were 4.2–5.2 points for Claude Sonnet 4.5 and diminished with stronger models"],"whyItMatters":"DeReAct shows how separating reasoning and acting can improve reliability in weaker AI agents without sacrificing grounding, offering a path to more trustworthy agent behavior.","category":{"slug":"agents","name":"Agents & Tools","url":"https://digestai.news/category/agents"},"entities":{"companies":[],"models":["Qwen3-Coder-480B","Claude Sonnet 4.5","Claude Opus 4.5"],"people":[]},"firstPublishedAt":"2026-10-05T04:00:00Z","updatedAt":"2026-10-05T04:00:00Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"arXiv cs.AI","title":"DeReAct: Decomposed Reasoning and Acting for Reliable AI Agents","url":"https://arxiv.org/abs/2610.02351","publishedAt":"2026-10-05T04:00:00Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"DeReAct improves weaker AI agents with external gating policies\", 5 October 2026, https://digestai.news/story/dereact-improves-weaker-ai-agents-with-external-gating-policies","publisher":"Digest AI","title":"DeReAct improves weaker AI agents with external gating policies","datePublished":"2026-10-05T04:00:00Z","url":"https://digestai.news/story/dereact-improves-weaker-ai-agents-with-external-gating-policies"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}