Reflexion: Language Agents with Verbal Reinforcement Learning

Summary

Reflexion gives agents a verbal self-feedback memory so failures can shape the next attempt without weight updates.

Why George Should Read It

It is a practical lens for lightweight agent improvement: preserve the lesson, feed it back into the next run, and verify behavior changes.

Best For

• agent loop design
• workflow memory
• evaluation-driven iteration

Notes

• Good companion to ReAct and later coding-agent loop papers.

Next Reading Action

Read the memory mechanism and write one analogous memory format for a real agent workflow.

Copyable Markdown

# Reflexion: Language Agents with Verbal Reinforcement Learning

Year: 2023
Category: Agents

Why it matters:
It is a practical lens for lightweight agent improvement: preserve the lesson, feed it back into the next run, and verify behavior changes.

Best for:
- agent loop design
- workflow memory
- evaluation-driven iteration

Next step:
Read the memory mechanism and write one analogous memory format for a real agent workflow.