Real-Time Procedural Learning From Experience for AI Agents

TL;DR

PRAXIS is a lightweight post-training mechanism enabling AI agents to learn procedural knowledge from experience in real time, improving task performance and adaptability in dynamic environments.

Contribution

The paper introduces PRAXIS, a novel method for real-time procedural learning that enhances AI agents' ability to acquire and recall task-specific knowledge after training.

Findings

PRAXIS improves task completion accuracy across different models.

PRAXIS increases reliability and cost efficiency in web browsing tasks.

Preliminary results show generalization to unseen tasks in similar environments.

Abstract

Learning how to do things from trial and error in real time is a hallmark of biological intelligence, yet most LLM-based agents lack mechanisms to acquire procedural knowledge after deployment. We propose Procedural Recall for Agents with eXperiences Indexed by State (PRAXIS), a lightweight post-training learning mechanism that stores the consequences of actions and retrieves them by jointly matching environmental and internal states of past episodes to the current state. PRAXIS augments agentic action selection with retrieved state-action-result exemplars that are generated in real time. When evaluated on the REAL web browsing benchmark, PRAXIS improves task completion accuracy, reliability, and cost efficiency across different foundation model backbones, and shows preliminary generalization to unseen tasks in similar environments. These results demonstrate that PRAXIS enables the practical adoption of AI agents in fast-evolving stateful environments by helping them learn new procedures effectively.