---
format: "aidr-story-markdown/v1"
id: "d647aa045876c2850b138d9ea2cb331bf3ed558c2e034a6b03ba23000c276636"
canonical_url: "https://aidr.today/d647aa04?lang=en"
title: "Procedural Graphs: Self-Evolving Execution Structures for LLM Agents"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-09-09T17:13:52.000Z"
category: "Research"
topics: ["agent","multi-agent","llm","google","reasoning","open-source","benchmark"]
source_urls: ["https://academy.dair.ai/papers/procedural-graphs-self-evolving-execution-structures-for-llm-agents-2609.09153","https://news.ycombinator.com/item?id=49629868","https://arxiv.org/abs/2609.09153"]
summary: "Yuxing Lu, Yicheng Chen, Shanchan Wu and Sercan Arik at Google, with Georgia Tech and Peking University, introduce the Procedural Graph, an explicit store of (p"
---

# Procedural Graphs: Self\-Evolving Execution Structures for LLM Agents

> [Open the canonical story](<https://aidr.today/d647aa04?lang=en>)

**Published:** 2026-09-09T17:13:52.000Z
**Category:** Research
**Topics:** agent, multi\-agent, llm, google, reasoning, open\-source, benchmark

## Summary

Yuxing Lu, Yicheng Chen, Shanchan Wu and Sercan Arik at Google, with Georgia Tech and Peking University, introduce the Procedural Graph, an explicit store of \(p

## Sources

- [Story source](<https://academy.dair.ai/papers/procedural-graphs-self-evolving-execution-structures-for-llm-agents-2609.09153>)
- [Discussion](<https://news.ycombinator.com/item?id=49629868>)
- [Story source](<https://arxiv.org/abs/2609.09153>)