Improving AI agents by combining quick updates with slow learning cycles
Experience Funnel: A State-Policy Alternating Loop for Self-Evolving Agents
Computation and Language
Summary
Many AI agents learn by interacting with their environment, but it’s hard to make their learning both fast and lasting. The authors propose a new approach called Experience Funnel, which helps agents update quickly with easy-to-understand notes about what they learn, then slowly turn the most useful knowledge into deeper skills. This back-and-forth process lets the agents get better over time, balancing quick changes with long-term improvement. Tests show this method helps agents improve more than just fast updates or slow learning alone.
Autonomous agentsLarge language modelsSelf-evolutionState adaptationPolicy consolidationExperience distillationInteraction trajectoriesTransition-aware learningModel competenceRollouts
Authors
Wenbo Gao, Zhaomou Song, Zhiyuan Ji, Renxi Liu, Xing Li, Xianzhi Yu, Xiaoguang Li, James Chung-wai Cheung, Weizhe Lin, Yaoyuan Wang
Abstract
Autonomous agents powered by large language models (LLMs) continuously accumulate experience through interaction, creating an opportunity to improve future behavior through self-evolution. A fundamental challenge is how to transform abundant, task-specific interaction experience into reusable model competence without sacrificing the ability to adapt rapidly to newly observed evidence. Explicit textual states, such as skills and agent harnesses, provide fast, human-readable and editable adaptation, but incur persistent dependence on external context; parametric policies provide compact and reusable competence, but are substantially slower to update. We present \textit{Experience Funnel}, a self-evolving framework that couples fast state adaptation with slow policy consolidation in an alternating loop. Interaction trajectories are first distilled into an explicit textual state, where newly acquired experience can be rapidly incorporated and validated. The framework then selectively identifies state-enabled behavior that remains useful across state revisions and consolidates it into the policy through transition-aware distillation. The updated state--policy pair subsequently generates new rollouts, providing fresh evidence for the next round of state adaptation and policy consolidation. Experiments across diverse agent benchmarks show that \textit{Experience Funnel} consistently improves agent capability over state-only evolution and policy-internalization approaches, while progressively converting useful explicit experience into autonomous policy competence.