PRIME: Training Free Proactive Reasoning via Iterative Memory Evolution for User-Centric Agent

TL;DR

PRIME introduces a gradient-free, experience-based framework for developing proactive, user-centric agents capable of iterative reasoning and communication in complex tasks, emphasizing cost-efficiency and interpretability.

Contribution

It presents PRIME, a novel gradient-free learning method that leverages structured experience accumulation and retrieval to improve agent performance without expensive training.

Findings

PRIME achieves competitive results with gradient-based methods.

It offers cost-efficient and interpretable agent training.

Experiments demonstrate effectiveness across diverse environments.

Abstract

The development of autonomous tool-use agents for complex, long-horizon tasks in collaboration with human users has become the frontier of agentic research. During multi-turn Human-AI interactions, the dynamic and uncertain nature of user demands poses a significant challenge; agents must not only invoke tools but also iteratively refine their understanding of user intent through effective communication. While recent advances in reinforcement learning offer a path to more capable tool-use agents, existing approaches require expensive training costs and struggle with turn-level credit assignment across extended interaction horizons. To this end, we introduce PRIME (Proactive Reasoning via Iterative Memory Evolution), a gradient-free learning framework that enables continuous agent evolvement through explicit experience accumulation rather than expensive parameter optimization. PRIME distills multi-turn interaction trajectories into structured, human-readable experiences organized across three semantic zones: successful strategies, failure patterns, and user preferences. These experiences evolve through meta-level operations and guide future agent behavior via retrieval-augmented generation. Our experiments across several diverse user-centric environments demonstrate that PRIME achieves competitive performance with gradient-based methods while offering cost-efficiency and interpretability. Together, PRIME presents a practical paradigm for building proactive, collaborative agents that learn from Human-AI interaction without the computational burden of gradient-based training.