AutoDesign: Meta-Harness Optimization for Long-Horizon Agentic Design

2026-08-13

AI Summary

Existing design systems struggle to align with human design priors and lack recursive self-improvement capabilities. This work proposes AutoDesign, a novel framework that introduces, for the first time, a recursively evolvable meta-harness mechanism. Guided by a meta-optimizer, code-based agents iteratively refine the design harness using execution feedback, enabling long-horizon autonomous design. By integrating human prior knowledge, AutoDesign transcends static design paradigms through multimodal structured generation and sustained tool orchestration. Evaluated on PosterBench, the method achieves a score of 78.32—outperforming Claude Design by 7.45 points (a relative improvement of 12.4%)—and receives the highest human preference in qualitative assessments. Moreover, it can generate conference-quality posters at low cost within 40 minutes.

Abstract

Transforming multimodal sources into condensed and structured media outputs can be fundamentally conceptualized as a long-horizon agentic process centered on a model-harness system. While an ideal harness system should align with human design priors and accumulate reusable experience through empirical exploration to drive recursive self-improvement, existing paradigms remain static and fall short of this capability. In this paper, we present AutoDesign, a framework that aligns with human design priors, where a meta-harness optimizer guides a code agent to recursively improve harness based on rollout feedback. To instantiate and evaluate this framework, we focus on the academic paper-to-poster generation task and introduce PosterBench, comprising a 100-paper Main Track spanning five disciplines and PosterBench-mini, a shared 10-paper subset for controlled evaluation. On the PosterBench Main Track, AutoDesign achieves the highest score of 78.32, surpassing the closed-source commercial system Claude Design by 7.45 points. Across seven controlled code-agent-model configurations, integrating the learned DesignHarness consistently improves performance, increasing the average PosterBench Score from 54.99 to 67.39 (+12.4%). In a fully autonomous long-horizon loop, it executes 253 tool calls and 11 editing turns within 40 minutes for under $3, reaching average conference-poster quality in human evaluation. A system-blind human study further demonstrates that AutoDesign achieves the highest human preference among evaluated systems.

Problem

• long-horizon agentic design
• model-harness system
• recursive self-improvement
• human design priors
• multimodal-to-structured generation

Innovation

• meta-harness optimization
• long-horizon agentic design
• recursive self-improvement
• human-aligned design priors
• code agent