OpenLoopEvolve: A Verifiable Self-Evolution Framework for Loop Policies in Long-Horizon Complex Tasks
2026-08-10 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors introduce OpenLoopEvolve (OLE), a system that helps agents learn and improve how they handle complex tasks over time by treating their problem-solving steps as reusable pieces called policies. OLE allows these policies to evolve either during task operation or by reviewing past experiences, using a method where new ideas are tested against old ones to find better ways. This approach helps agents get better at planning, acting, and recovering from mistakes in changing situations. Tests showed that OLE improves performance compared to using a single fixed strategy.
Long-horizon tasksLoop PolicyPolicy evolutionLarge language modelsOnline learningOffline learningAgent planningTask performanceAutomated recoveryBenchmark evaluation
Authors
Siqi Wang, Xinlin Li, Zhenglin Li, Li Li
Abstract
Long-horizon complex tasks require agents to repeatedly observe states, formulate plans, invoke tools, verify results, and recover from failures in continuously changing environments. However, such control experience often remains confined to a single context or a fixed prompt, and is difficult to accumulate and reuse across historical traces. This paper presents OpenLoopEvolve (OLE), a self-evolution framework centered on the Loop Policy. OLE represents an agent's observation, planning, memory, action, verification, recovery, stopping, and budget control as portable policy assets with versions and lineages, and provides online and offline evolution modes that can be selected according to practical needs: the online mode triggers candidate generation based on feedback from continuous operation, whereas the offline mode searches for candidate policies from archived traces and failure evidence. Both modes share an evolution mechanism consisting of autonomous proposals by a large language model, Champion--Challenger paired evaluation, and robust release. Policies released online are activated at a subsequent task boundary, monitored using subsequent feedback, and rolled back to their parent versions when degradation conditions are met. On the simulated business benchmark YC-Bench, both modes improve aggregate task performance, task success rate, and risk metrics relative to a fixed initial Loop Policy. The results indicate that treating the Loop Policy as a governable asset can support the accumulation, comparison, release, and reuse of control experience and improve agent performance on long-horizon complex tasks.