Escher-Loop: Mutual Evolution by Closed-Loop Self-Referential Optimization

TL;DR

Escher-Loop introduces a closed-loop framework for mutual evolution of task and optimizer agents, enabling continuous self-improvement and surpassing static baseline performance in optimization tasks.

Contribution

The paper presents a novel self-referential system that dynamically evolves task and optimizer agents using a benchmarking mechanism based on empirical scores.

Findings

Achieves highest performance across evaluated optimization tasks.

Optimizer agents adapt strategies to match high-performing task agents.

System demonstrates continuous improvement beyond static baselines.

Abstract

While recent autonomous agents demonstrate impressive capabilities, they predominantly rely on manually scripted workflows and handcrafted heuristics, inherently limiting their potential for open-ended improvement. To address this, we propose Escher-Loop, a fully closed-loop framework that operationalizes the mutual evolution of two distinct populations: Task Agents that solve concrete problems, and Optimizer Agents that recursively refine both the task agents and themselves. To sustain this self-referential evolution, we propose a dynamic benchmarking mechanism that seamlessly reuses the empirical scores of newly generated task agents as relative win-loss signals to update optimizers' scores. This mechanism leverages the evolution of task agents as an inherent signal to drive the evaluation and refinement of optimizers without additional overhead. Empirical evaluations on mathematical optimization problems demonstrate that Escher-Loop effectively pushes past the performance ceilings of static baselines, achieving the highest absolute peak performance across all evaluated tasks under matched compute. Remarkably, we observe that the optimizer agents dynamically adapt their strategies to match the shifting demands of high-performing task agents, which explains the system's continuous improvement and superior late-stage performance.