ICLR 2025: Automated Design of Agentic Systems
blaizedsouza · x · 2026-07-19
This ICLR 2025 paper introduces Automated Design of Agentic Systems (ADAS), treating agents as code to be searched rather than relying on manually designed prompts and workflows.
The core workflow involves:
- A meta agent generating new agent programs;
- Executing and scoring them on tasks;
- Archiving better-performing agents;
- Iteratively searching for superior designs.
The authors emphasize that this search space covers prompts, tool calls, control flows, and their combinations, theoretically uncovering novel agent architectures beyond human design. Experiments across coding, science, and math show the method progressively discovers stronger agent designs while maintaining solid generalization in cross-domain and cross-model transfers.
More from coding & agent
- A roundup of AI agents and MCP resources, including how to evaluate agents — _jaydeepkarale · 2026-07-21
- A full course shows how to build and deploy an AI agent with OpenAI and LangChain — _jaydeepkarale · 2026-07-21
- A beginner guide to AI agents points readers to a Stanford webinar — _jaydeepkarale · 2026-07-21
- A practical guide on how to evaluate AI agents — _jaydeepkarale · 2026-07-21
- MCP is headed toward easier scale, event-driven extensions, and workable file uploads — EricBuess · 2026-07-21
- Developers debate the missing composition model for AI agents — threepointone · 2026-07-21