The paper introduces AutoDesign, a framework that employs a meta-harness optimizer to guide a code agent in recursively improving its harness based on rollout feedback. The authors evaluate this approach on the academic paper-to-poster generation task using a new benchmark called PosterBench.
- AutoDesign achieves a score of 78.32 on the PosterBench Main Track, surpassing Claude Design by 7.45 points.
- Integrating the learned DesignHarness increases the average PosterBench Score from 54.99 to 67.39 across seven code-agent-model configurations.
- In a fully autonomous loop, the system executes 253 tool calls and 11 editing turns within 40 minutes for under $3.
- A system-blind human study shows AutoDesign achieves the highest human preference among evaluated systems.
The authors consider this important because it demonstrates how aligning with human design priors and accumulating reusable experience can drive recursive self-improvement in long-horizon agentic processes.