Jay Lee

I am interested in AI memory that helps people think better: turning transcripts, notes, and everyday context into insights they can inspect, question, and use.

Work

The first object here is the memory contract: AI should not imitate a person. It should help the person notice, compare, and decide better.

Memory systemAI Memory Contract

A note pattern for AI memory that helps judgment instead of imitating the person.

Agent workflowVoice Memo Decision Notes

A local audio workflow that turns transcripts into decisions, risks, and next questions without publishing the room.

Public repoAI Workflow Audit Kit

A public repo for turning messy workflow descriptions into inspectable AI experiment notes and eval gates.

PrototypeLoop Lab

A prototype direction for collect, draft, evaluate, and revise loops around everyday knowledge work.

Small Tools

Open-source utilities from the same workbench. Small, local-first, and inspectable.

Open sourcephoto-trip-timeline

A local-first Apple Photos utility that turns trips into a shareable Markdown timeline without exporting GPS data.

Open sourceclaude-voice-memo

A voice recording workflow for speaker-aware transcripts and key briefings using local transcription tools.

Open sourcedopamind

An ADHD-friendly learning tool that turns long content into focused chunks, hooks, and catch-me-up recaps.

Notes

Short field notes tied to the public work. No private transcripts, employer material, or personal records.

MemoryAI memory needs a human-auditable contract

Memory should surface sources, boundaries, and better questions, not become hidden personality capture.

AudioTurning voice memos into decision notes without leaking the room

Local transcripts can become decisions, risks, and follow-ups without exposing raw conversations.

Build noteFrom workflow descriptions to inspectable AI experiments

Why the audit kit starts with owners, baselines, risk boundaries, and eval evidence.

Now

  • designing AI memory that helps people notice, compare, and decide better
  • turning meeting transcripts into insight without publishing the room
  • keeping source notes reviewable before an agent is allowed to reuse them
Open Now

한국어 요약

이 사이트는 AI가 사람을 흉내 내는 방향이 아니라, 사람이 더 잘 판단하도록 돕는 기억과 작업 시스템을 다루는 개인 아카이브입니다.

한국어 요약 보기