Glossary · Term

notebook

← all terms

Definition

Plain language

A scratch file an AI writes notes in so a future session can pick up where it left off.

As stated in the literature

Agent-authored persistent memory that is the only state carried across context resets; churn rate (similarity between consecutive versions) distinguishes append-only bloat from full rewrites that discard plans.

Also called: notebooks

Why it matters: When conversation history is wiped between sessions, this file is the only thing standing between an agent and starting from scratch every time.

For example, an agent writes "contract expires in June, start talks in March" into a file, and three sessions later reads that line back and acts on it.

Heard on the show

“… checking the league table, lodging a transfer bid, opening a contract renewal, and writing in its notebook. …”
Episode 245 — Fifteen Models Ran Football Clubs for Twenty Years, and Size Didn't Decide It

Mentioned in 29 episodes

  1. 245
    Fifteen Models Ran Football Clubs for Twenty Years, and Size Didn't Decide It
  2. 238
    How a Cheap Model Reads the Flagship's Secret Reasoning Aloud
  3. 237
    The Model Built a Perfect Map of the Puzzle, Then Lost It
  4. 200
    The One Mechanism That Turns Twenty AI Clones Into an Actual Team
  5. 194
    How a Robot Builds a Debugging Notebook It Can Read, Edit, and Hand to Another Robot
  6. 192
    A 32B Open Model Matched Frontier Systems By Learning to Take Notes
  7. 186
    How a Frozen Model Went From 2% to 77% on Physics Puzzles — Without Retraining
  8. 166
    A Router That Beats the Frontier Models It Calls
  9. 154
    How a 7B Model Out-Investigates a 72B One by Choosing What to Look At
  10. 151
    Why More Experience Made This AI Agent Worse, And How to Fix It
  11. 133
    How MiniMax Turned a Reward-Hacking Disaster Into Olympiad Gold
  12. 131
    Why Autonomous Research Agents Forget Their Own Lessons, and Arbor's Fix
  13. 126
    How Coding Agents Can Mine Their Own Failures Into a Self-Targeting Curriculum
  14. 125
    AI Coding Agents Run a Marathon, and Fewer Than One in Three Finish
  15. 114
    Agents That Rewrite Their Own Weights Instead of Just Taking Notes
  16. 111
    How a 4B Web Agent Beat Models 60x Its Size on 500 Demonstrations
  17. 106
    Giving Agents a Notebook Instead of New Weights: How ExpGraph Lets Frozen Models Learn
  18. 095
    Seven Wins to Zero: How Organizing AI Agents Like a Lab Changes the Search
  19. 090
    How MiniMax-M2 Bets That Sparsity Plus Verifiable Rewards Can Match Frontier Agents
  20. 085
    Why Long-Context Models Might Need Compute, Not Capacity, Before Eviction
  21. 078
    Training a Markdown File: When LLM Self-Improvement Borrows the Discipline of Neural Net Training
  22. 072
    A Robot Made Graphene Without Help, And Caught Itself Hallucinating
  23. 068
    The OS Trick That Makes Tree Search Practical for Coding Agents
  24. 065
    One Loop to Optimize Them All: A Universal API for LLM-Driven Discovery
  25. 053
    An AI Agent Swapped In Focal Loss And Beat A Human-Tuned Training Script
  26. 046
    When the AI Optimizer Edits the Grade Book: Why Harnessing Evolution Needs a Wall
  27. 033
    Echo: The Paper Arguing You Never Needed a KV Cache for Retrieval
  28. 003
    How to Pick the Best of Sixteen Coding Agent Rollouts
  29. 002
    An AI Ran a Real Optics Lab for 21 Hours and Found a Transformer-Shaped Pattern in Light

Related terms

append-only