When an assistant forgets, there is almost always a specific reason: a setting, a limit, a scope or a full context window. These posts find the reason and fix it, step by step.
open the folder →17 sheets on the desk · updated Nov 18, 2026
The Esment Blognotes on AI memory
Fixes for when ChatGPT or Claude forgets. Guides to Projects, imports and second brains. How coding agents get their context. And lab notes from the team building Esment, the memory your assistants share.
pick a pencil
- Memory fixes — when ChatGPT or Claude forgets, the reason and the fix, step by step
- Guides — hands-on: imports, Projects and second brains
- Agents & context — AGENTS.md, hooks and what reaches a coding agent’s context
- Esment Lab — our experiments, with the numbers — including the misses
- Essays — longer arguments about how assistant memory should work
On the desk
— start hereJust drawn
— the newest sheetsAI second brain: what it is, and how to build one that remembers
ChatGPT memory full? A cleanup routine you can repeat
Claude Code hooks for memory: load a brief, save notes, survive compaction
Claude memory not working? A checklist for the apps and Claude Code
The shelves
— everything, by kindLonger, hands-on guides for people who use AI assistants every day: how to move your history between them, how Projects and memory fit together, and how to build a second brain an assistant can actually use.
open the folder →Coding agents only know what reaches their context window. These posts cover the files, hooks and habits that decide what gets there — and what a persistent memory adds on top.
open the folder →The Lab is where Esment gets measured before it ships. These notes report the experiments as they ran — the plan, the numbers, the failures — including the changes we decided not to ship.
open the folder →Longer arguments about how assistant memory should work: what it is for, where it should live, and how it should grow, merge and forget.
open the folder →The index
— every sheet, newest first- 17AGENTS.md vs CLAUDE.md: what it is, who reads it, and how to use bothAgents & context · Nov 18, 2026Agents & context
- 16AI second brain: what it is, and how to build one that remembersGuides · Nov 18, 2026Guides
- 15ChatGPT memory full? A cleanup routine you can repeatMemory fixes · Nov 18, 2026Memory fixes
- 14ChatGPT memory not working? A checklist to find the causeMemory fixes · Nov 18, 2026Memory fixes
- 13Claude Code hooks for memory: load a brief, save notes, survive compactionAgents & context · Nov 18, 2026Agents & context
- 12Claude memory not working? A checklist for the apps and Claude CodeMemory fixes · Nov 18, 2026Memory fixes
- 11Context engineering: what it is, where memory fits, and a practical checklistAgents & context · Nov 18, 2026Agents & context
- 10How memory works in ChatGPT Projects and Claude ProjectsGuides · Nov 18, 2026Guides
- 09How to import your ChatGPT chat history (and what actually transfers)Guides · Nov 18, 2026Guides
- 08Multilingual memory search: how we tested it and what we changedEsment Lab · Nov 18, 2026Esment Lab
- 07Second brain apps in 2026: what to pick if you use AI every dayGuides · Nov 18, 2026Guides
- 06What each stage of our memory retrieval cascade is worthEsment Lab · Nov 18, 2026Esment Lab
- 05Why AI memory answers fail: retrieval or reading?Esment Lab · Nov 18, 2026Esment Lab
- 04Why ChatGPT forgets, and how to tell which kind of forgetting you haveMemory fixes · Nov 18, 2026Memory fixes
- 03Why AI assistants need a memory they can't loseEssays · Aug 10, 2026Essays
- 02Local-first AI memory is a feature, not a constraintEssays · Jul 28, 2026Essays
- 01What we learned building a temporal knowledge graph for AI memoryEsment Lab · Jul 2, 2026Esment Lab
Hashtags
— the threads between sheetsa note for you
New posts, by email.
Fixes, guides and Lab notes on AI memory — about two a month. No product news, unsubscribe in one click.