---
title: "Captured: Claude — Memory Layer Discussion"
url: https://memory.wiki/ycd29v7A
updated: 2026-05-09T21:14:14.503Z
hub: https://memory.wiki/hub/yc-demo
concept_count: 10
source: "memory.wiki"
---
# Captured: Claude — Memory Layer Discussion

> **You**: I'm thinking about authored memory vs extracted memory.
> Where does extracted memory fail?

> **Claude**: Three places, in order of severity:
>
> 1. **Misextraction** — the model picks up a passing comment as a
>    "preference" and re-inserts it forever. Hard to undo because
>    the user doesn't know it's there.
>
> 2. **Walled gardens** — extracted memory lives in the vendor's
>    DB. Switch vendors → memory doesn't follow. Lock-in by
>    accident.
>
> 3. **Audit trail** — extracted memory has no edit log the user
>    can reason about. Authored memory has commits, diffs, version
>    history.

> **You**: So when does extracted win?

> **Claude**: Casual personal use. "Remember that I'm vegetarian"
> doesn't need a versioning system. Authored memory wins for
> high-stakes, multi-AI, or long-lived knowledge.


---

## Concepts in this document
- **Vendor lock-in** _(concept)_
  The central argument that proprietary memory systems create strategic dependency on a single platform.
- **Authored memory** _(concept)_
  User-controlled memory with explicit commits and version history, preferred for high-stakes and multi-AI scenarios.
- **Extracted memory** _(concept)_
  Automatically inferred memory from conversational context, prone to misextraction and lacking user transparency.
- **Misextraction** _(concept)_
  Critical failure mode where passing comments are incorrectly captured as persistent preferences without user awareness.
- **Audit trail** _(concept)_
  Edit history and versioning capability that allows users to understand and reason about memory changes.
- **Casual personal use** _(concept)_
  Low-stakes, simple preference storage where extracted memory's simplicity provides sufficient value.
- **Memory layer** _(concept)_
  The system component responsible for persisting and retrieving user preferences and knowledge across AI interactions.
- **Portability** _(concept)_
  Ability to transfer memory across different AI systems and vendors without loss of information.
- **Multi-AI workflows** _(concept)_
  Scenarios requiring memory portability across different AI systems, where authored memory provides better compatibility.
- **High-stakes knowledge** _(tag)_
  Mission-critical or long-lived memory that benefits from authored structure, versioning, and multi-AI coordination.

## Concept relations (within this doc's concepts)
- **Extracted memory** vulnerable to **Misextraction**
- **Extracted memory** creates risk of **Vendor lock-in**
- **Authored memory** provides **Audit trail**
- **Authored memory** enables **Portability**
- **Extracted memory** suited for **Casual personal use**
- **Authored memory** better for **High-stakes knowledge**
- **Authored memory** enables portability in **Multi-AI workflows**
- **Extracted memory** suitable for **Casual personal use**
- **Authored memory** architecture approach for **Memory layer**
- **Authored memory** includes **Audit trail**

_Hub canonical:_ https://memory.wiki/hub/yc-demo
_Concept digest:_ https://memory.wiki/raw/hub/yc-demo?digest=1&compact=1