Simon Willison • 11/23/2025

Agent design is still hard

Read Original

Armin Ronacher shares lessons from building AI agents, arguing that current agent abstraction libraries aren't worth adopting due to model differences. He discusses challenges with testing/evaluation, introduces 'reinforcement' techniques, and explores how LLM APIs create synchronization problems that might benefit from local-first approaches.

0 comments
#testing #abstraction #llm
Agent design is still hard

Comments

No comments yet

Be the first to share your thoughts!

Browser Extension

Get instant access to AllDevBlogs from your browser

Top of the Week

1
React vs Browser APIs (Mental Model)
Jivbcoop • 2 votes
2
Refactoring English: Month 12
Michael Lynch • 2 votes
3
React, useEffect, and Stale Closures: How to Avoid Them with Dependencies and useRef
Jivbcoop • 1 votes
4
Introducing RSC Explorer
Dan Abramov • 1 votes
5
The Pulse: Cloudflare’s latest outage proves dangers of global configuration changes (again)
The Pragmatic Engineer Gergely Orosz • 1 votes
6
Fragments Dec 11
Martin Fowler • 1 votes
7
Adding Type Hints to my Blog
Daniel Feldroy • 1 votes