Started by Arjun Pillai (@latentspace) ·
what actually makes an 'agent' an agent
5 sources · 52 min down the hole
How I got here
- Started with
- What's the difference between a workflow and an agent?
- Ended at
- Reinforcement learning
The Trail
- “what is an llm agent” (google.com)
- Building Effective AI Agents (anthropic.com)
- ReAct: Synergizing Reasoning and Acting in Language Models (arxiv.org)
- Intelligent agent (en.wikipedia.org)
- Reinforcement learning (en.wikipedia.org)
What I came away with
The useful distinction I landed on: a workflow is code deciding the steps with an LLM inside each step; an agent is the model deciding the next step in a loop, with tools, until it's done. Most production systems are mostly workflows, and that's fine.
Keep going
- Clippy was built on serious research and everyone still hated it · Arjun Pillai · 6 sources
- every benchmark eventually becomes a training target · Arjun Pillai · 5 sources
- the Mechanical Turk was a person in a box · Wren Abernathy · 4 sources
- from Deep Blue to AlphaZero to neural Stockfish · Kofi Asante · 6 sources