Google Paper: Editable Procedural Graphs Rank 1st in 21 of 24 Agent Benchmarks
rohanpaul_ai · x · 2026-09-21
A new Google paper introduces Procedural Graphs: keeping an LLM agent's long-task workflow in an editable procedure graph instead of burying it in chat history.
Problem: agents running long tasks forget where they are, repeat tool calls, or execute steps out of order.
Method:
- The graph gives the agent a small map of possible next steps while leaving the LLM free to reason
- After runs, another LLM compares successes and failures and edits the map — edits are kept only if they don't hurt held-out tasks
Results:
- Ranked 1st or tied 1st in 21 of 24 model-benchmark combinations
- Repaired a flawed human-designed workflow: MultiChallenge success went from 58.93% to 92.86% after refinement
Caveats: the guidance costs extra tokens, so it fits cases where long workflows are the bottleneck. Core recommendation: move critical procedures out of chat history into an explicit, execution-improving workflow.
Related event: Google's Procedural Graphs boost LLM agent long-task performance(2 posts)→
More from coding & agent
- Agent seller confesses: his 'flawless' four-month automation needed twice-weekly secret patches — Cold_Mud2650 · 2026-09-21
- Free DiffusionGemma endpoint returns sentence probabilities in ~230ms, OpenAI-compatible — victormustar · 2026-09-21
- Mozilla AI runs a local 30B model end-to-end to open a real bugfix PR, fully offline — mozilla-ai · 2026-09-21
- Dev demo: agent codes a game and generates all 3D assets itself via Hyper3D Rodin MCP — CodeByPoonam · 2026-09-21
- mcp-server-dns: Zero-config MCP server for DNS, WHOIS and domain checks — modelcontextprotocol · 2026-09-21
- Each MCP tool passes solo tests, but the agent's full workflow breaks on state passing — Stock-Pumpkin-8859 · 2026-09-21