theNotebook

Topic

Agents and tool use

Agents that plan, call tools and act. What breaks between the model and the world.

12 daily entries · 4 concepts · 1 paper

Kitchen table

An agent is a helper that can press buttons for you. It plans a step, presses a button and looks at what happened. These pages are about the times it presses the wrong one.

Whiteboard

An agent is a model plus a loop plus tools plus state. Most failures sit in the loop and the tools, not the model. These entries cover retries, side effects, skills and the harness.

Boardroom

An agent acts on your systems, not just your screen. Every action has a cost when it goes wrong. These pages show where the control points belong.

Start with the concepts

  • AI Agents A model plus a loop plus tools plus state, and the model is the smallest part.

  • Loops & Graphs A loop decides at runtime – model in charge, maximum flexibility, unbounded variance.

  • Tools & MCP Function calling with schemas, and MCP emerging as the standard for exposing tools across vendors – write once, plug into any model.

  • Memory Three different problems wearing one name: working context (this session), episodic notes (what happened before), durable facts (what stays true).

Papers

From the daily page

Search the notebook