Self-improving AI / ADAS: Automated Design of Agentic Systems
ADAS: Automated Design of Agentic Systems
“We present a simple yet effective algorithm named Meta Agent Search to demonstrate this idea, where a meta agent iteratively programs interesting new agents based on an ever-growing archive of previous discoveries.”
ADAS: Automated Design of Agentic Systems is one of the projects on the self-improving AI map. Below is what it changes about itself, whether the change lasts, and whether it improves the improver, each with the line from its own source.
Where it sits
Columns: what it changes, from the surface to the deepest change. Rows: whether the change outlasts the run. A number is how many projects sit in that square. See it on the 3D map
- What it changes
- Prompts, memory, context“add them to an archive of discovered agents, and use this archive to inform the meta agent in subsequent iterations”
- Does the change last
- Gone when the run ends“the iteration continues with the updated archive until the maximum number of iterations is reached”
- Does it improve the improver
- does not improve how it improves“Meta Agent Search runs for 25 iterations and the meta agent uses GPT-4”
- Who sets the goals
- people set tasks and scoring“Here, we calculate the performance (e.g., success rate or F1 score) and 95% bootstrap confidence interval as the metrics for the meta agent to maximize.”
- How it was tested
- tested on held-out tasks“We report the test accuracy and the 95% bootstrap confidence interval on held-out test sets.”
What grows is an archive of agent designs written as Python code; the paper says searching in code theoretically enables any agentic system, including prompts, tool use and control flows. The meta agent that writes them is GPT-4, and the discovered agents run on GPT-3.5; the paper lists letting the meta agent itself be improved through ADAS as a future direction. Humans choose each domain and the validation score the meta agent maximizes; discovered agents were scored on held-out test sets and transferred to other domains and models.
Does it count as RSI?
It depends on the definition. Under the four used on What counts as RSI, by what its own source says:
- Loose (Eliezer Yudkowsky, Intelligence Explosion Microeconomics (2013)): does not count
- Durable (Ren et al., self-improvement survey (2026)): does not count
- Strict (Roman Yampolskiy, From Seed AI to Technological Singularity (2015)): does not count
- Autonomous (Anthropic, When AI builds itself): does not count
How it got here
Earlier work in this catalog that ADAS cites, oldest first. Under each: where that work sits on the map, then where ADAS sits (changes in bold), and what ADAS's own paper says about it. Its whole family on the citation map
2023 Reflexion uses it
Prompts, memory · gone when the run ends · does not improve the improver
“We also adopt self-reflection ([Shinn et al., 2023](https://arxiv.org/html/2408.08435v1#bib.bib76); [Madaan et al., 2024](https://arxiv.org/html/2408.08435v1#bib.bib50)) iterations in our meta agent, where it performs two iterations of refinement on the novelty and correctness of the proposal and performs up to three refinements when errors occur while running the code.”
2023 Self-Refine uses it
Prompts, memory · gone when the run ends · does not improve the improver
“We also adopt self-reflection ([Shinn et al., 2023](https://arxiv.org/html/2408.08435v1#bib.bib76); [Madaan et al., 2024](https://arxiv.org/html/2408.08435v1#bib.bib50)) iterations in our meta agent, where it performs two iterations of refinement on the novelty and correctness of the proposal and performs up to three refinements when errors occur while running the code.”
2023 Promptbreeder sets itself apart
Prompts, memory · gone when the run ends · improves the task agent → does not improve the improver
“For example, works like PromptBreeder ([Fernando et al., 2024](https://arxiv.org/html/2408.08435v1#bib.bib26)) mutate only the text prompts of an agent, but their other components, such as control flow, remain the same.”
2024 ADAS
Prompts, memory · gone when the run ends · does not improve the improver
Also cites Voyager 2023, The AI Scientist 2024
Cited by Gödel Agent 2024, SICA 2025, Darwin Gödel Machine 2025, Hyperagents 2026, Ren, self-improvement 2026, Weco AIDE² 2026, Recursive Harness Self-Improvement 2026, Fibona Agent 2026, Harness² 2026, Self-HPO 2026, RSI-Opt 2026, Skill∞ 2026
In the same square
AlphaEvolve, AREX-2, LM2LM, MLEvolve, ParetoRSI-Gym, PostTrainBench, Promptbreeder, Reflexion, Self-Refine, The AI Scientist