Procedural graphs improve large language model agents planning and acting
Procedural Graphs: Self-Evolving Execution Structures for LLM Agents
Artificial IntelligenceComputation and LanguageMultiagent Systems
Summary
When large language models (LLMs) try to plan and carry out complex tasks, they often forget what to do next or repeat mistakes over long steps. The authors introduce a way to structure these tasks using "procedural graphs," which organize what actions to take and when, like a map for the model's decisions. This graph can update itself by learning from past successes and failures, improving over time without needing humans to fix it. Their experiments show this approach helps models work more reliably and efficiently on various tasks.
What this means in practice
- •For software engineers: Create adaptive AI agents that plan and act in steps more reliably by using self-updating procedural graphs to guide decisions over long tasks.
- •For automation system developers: Improve task automation by structuring procedural knowledge in graphs that evolve from experience, reducing repeated errors and maintaining goal focus.
Authors
Yuxing Lu, Yicheng Chen, Shanchan Wu, Sercan Ö. Arık
Abstract
Large language models are increasingly deployed as agents that plan over long horizons and act through external tools. Most agents select actions through unconstrained generation over an accumulating history, leaving implicit the procedural knowledge of what to do, in what order, and under which conditions. As trajectories lengthen, agents can lose track of their objectives, invoke tools out of order, and repeat unproductive actions. We introduce the Procedural Graph: just as a knowledge graph organizes factual knowledge into (entity, relation, entity) triplets for what-is questions, a Procedural Graph organizes procedural knowledge into (procedure, relation, procedure) triplets for what-to-do questions. At each decision step, the framework localizes the agent's active node, and a guidance model translates the surrounding subgraph into step-level situational guidance that biases the solver's next action without dictating it. The graph is self-evolving: an LLM refiner contrasts failed trajectories with successful ones and edits the graph's topology and attributes, committing edits that preserve or improve held-out validation performance while retaining rejected ones to discourage repetition. Starting from a minimal skeleton, the loop builds graphs that match or surpass hand-designed ones. It can also repair a flawed expert prior. Across multiple datasets, task types, and LLMs, the Procedural Graph delivers consistent gains over memory-based baselines, and self-evolution further improves performance without manual engineering.