AutoAgent: Evolving Cognition and Elastic Memory Orchestration for Adaptive Agents

Researchers introduce AutoAgent, a self-evolving multi-agent framework that combines evolving cognition, contextual decision-making, and elastic memory orchestration to enable adaptive autonomous agents. The system continuously learns from experience without external retraining and shows improved performance across retrieval, tool-use, and collaborative tasks compared to static baselines.

• →AutoAgent addresses limitations of current autonomous agent frameworks including static cognition and rigid workflow dependence.
• →The framework features three core components: evolving cognition, on-the-fly contextual decision-making, and elastic memory orchestration.
• →Agents maintain structured prompt-level cognition over tools, capabilities, peer expertise, and task knowledge that evolves through experience.
• →The Elastic Memory Orchestrator reduces token overhead while preserving decision-critical information through compression and abstraction.
• →Empirical testing shows consistent improvements in task success, tool-use efficiency, and collaborative robustness across multiple benchmarks.