Simple Agent Lab

Simple Agent Lab is an open AI research collective. We’re building a future where AI can help build AI, while the methods behind it remain understandable to humans. Simple Agent Lab 是一个开放的 AI 研究共同体。我们希望构建一个由 AI 帮助创造、又能被人类理解的未来。

AI will increasingly write code, generate data, run experiments, evaluate results, and improve models and systems. Improved AI can then return to build the next generation—forming a recursive loop of improvement. AI 将越来越多地编写代码、生成数据、运行实验、评估结果，并改进模型与系统。改进后的 AI 可以继续参与下一轮构建，形成递归的改进闭环。

But a future built by AI should not become too complex for humans to understand. We want every improvement to remain simple, transparent, and verifiable—so people can understand how it happened and decide where it goes next. 但由 AI 构建的未来，不应复杂到无法被人类理解。我们希望每一次改进都保持简单、透明、可验证，让人们能够理解它如何发生，并决定它走向何方。

From AI4AI to recursive improvement 从 AI4AI 走向递归改进

We research AI4AI and Self-Evolving Agents, build Self-Improving Systems and Self-Improving Skills, and explore the path toward Recursive Self-Improvement (RSI). 我们研究 AI4AI 与 Self-Evolving Agents（自进化智能体），构建 Self-Improving Systems（自改进系统）与 Self-Improving Skills（自改进技能），并探索通往 Recursive Self-Improvement（RSI，递归自我改进） 的路径。

Simple yet effective Simple yet effective

Simple does not mean unsophisticated. Simplicity is what allows humans to continue understanding and participating in a future built by AI. We want people to see how a method works, verify why an improvement happened, and help decide what comes next. 简单不代表简陋，简单是让人类始终能够理解并参与 AI 构建未来的前提。我们希望人们能够看懂方法如何工作，验证改进为何发生，并参与决定下一步。

Research is better when shared 研究因分享而进步

We share code, experiments, evaluations, and failures so others can inspect, reproduce, and build on what we learn. Our members have studied or worked at ByteDance, UC Berkeley, Tsinghua University, Shanghai Jiao Tong University, and Tongji University. All members participate in a personal capacity. 我们分享代码、实验、评测与失败，让更多人能够检查、复现并继续推进这些探索。我们的成员曾在 ByteDance、UC Berkeley、清华大学、上海交通大学和同济大学学习或工作。所有成员均以个人身份参与。