The continuity layer for human thinking

Your thinking follows you.

AI conversations are ephemeral.
Thinking is cumulative.

Thread remembers the ideas you develop with AI — how they evolve and where you left off.

Press ⌘ ⇧ T to recall — from any app.

  • Claude
  • ChatGPT
  • Gemini
  • DeepSeek
  • Cursor

Computable Authority

Institutional authority should be executable — and independently verifiable.

Evolved across 3 conversations

  1. Aug 17

    ChatGPT

    Authority needs explicit boundaries.

  2. Aug 19

    Cursor

    Those boundaries need to be executable.

  3. Aug 23

    Claude

    Execution isn't enough. Authority must also be independently verifiable.

Open question

Who performs the independent verification?

A thought, carried forward.

An idea can evolve across ChatGPT, Cursor and Claude without getting lost.

Don't lose the thought just because the chat ended.

You can have a hundred conversations with AI and still lose the one idea that mattered.

Ideas move between chats and change over weeks. Often you only remember the thought — not where you had it.

Thread is built to remember that.

01IDEAS GET BURIED

Your thinking isn't linear. Your chats are.

The important idea might be 200 messages deep in a conversation you haven't opened in three weeks.

02SEARCH ISN'T ENOUGH

You can't search for what you can't remember.

You remember what you were thinking about, not necessarily the words you used.

03CONTEXT GETS RESET

A new chat often means reconstructing the reasoning all over again.

The work you did yesterday should be able to come with you today.

The shift

Don't remember the chat.

Remember the thought.

Thread turns conversations into a persistent representation of the ideas you're actually developing.

ConversationIdeaEvolutionCurrent stateContinue

Your ideas have a history.

An idea isn't a static note. It emerges. It changes. It connects. It gets challenged. It becomes a decision. Or it stays unresolved.

Thread captures that evolution.

Computable Authority

ChatGPT → Cursor → Claude · Aug 17–23

Developing · Idea

Current formulation

Institutional authority should be expressed in a form an AI can execute — and another party can independently verify.

Evolution · 3 conversations

  1. Aug 17

    ChatGPT

    Authority needs explicit boundaries.

  2. Aug 19

    Cursor

    Those boundaries need to be executable.

  3. Aug 23

    Claude

    Execution isn't enough. Authority must also be independently verifiable.

Open loop

Who performs the independent verification?

Related: NOMOS · Policy-as-Code · AI Governance

Talk normally. Keep thinking.

You don't need to change how you work.

Use ChatGPT.

Use Claude.

Use Gemini.

Use Cursor.

Use whatever comes next.

Thread works underneath them.

01

Talk

Use AI normally.

02

Capture

Thread identifies meaningful thinking.

03

Connect

New thinking connects to ideas you've developed before.

04

Evolve

Ideas are refined rather than duplicated.

05

Recover

Find the thought, not the chat.

06

Continue

Pick up where you left off.

"Where was I with this?"

You

Where was I with independent verification?

Thread

Independent Verification of Authority

You first explored this on August 17 while developing Computable Authority.

Your thinking evolved from — across 3 conversations

  1. 01
    ChatGPT

    Authority needs explicit boundaries.

  2. 02
    Cursor

    Those boundaries need to be machine-executable.

  3. 03
    Claude

    Execution alone isn't sufficient.

  4. 04
    Claude

    A separate party should be able to verify that an action conforms to the authority.

Current open question

Who performs the verification, and what exactly should they verify?

Related: Computable Authority · NOMOS · Policy-as-Code

And your AI can pick it up too.

Thread doesn't just help you recover your thinking. It can give that thinking back to the AI you're working with.

The AI doesn't need your entire history. It gets the thinking that matters.

Under the hood, this happens through MCP and APIs — compatible AI tools access the Thinking State directly.

Thread

Thinking State

ChatGPT · Claude · Gemini

Continue

Your thinking shouldn't belong to one model.

ChatGPT can remember things about you. So can Claude.

But that memory doesn't travel between them.

ChatGPT is one place you think. Claude is another. Gemini is another.

Your models can change. Your tools can change. Your thinking shouldn't have to start over.

One thinking layer. Every AI.

Thread sits across the AI tools you already use. Conversations from different sources feed a shared Thinking State — one persistent representation of the ideas you're developing.

You can recover it yourself, or compatible AI tools can access the relevant parts when you need them.

CHATGPTCLAUDEGEMINICURSORVOICETHREADTHINKING ENGINEIDEA NODES01EVOLUTION02OPEN LOOPS03DECISIONS04THINKING STATEYOUANY AIFIG. 01 — MODEL-INDEPENDENT CONTINUITYTHREAD / LAYER

AI made conversation abundant.

Thread makes thinking cumulative.

The value of AI isn't only the answers it gives you. It's what happens to your thinking between them.

Free for 14 days. Then Pro.

No card for the trial. Everything you capture stays recoverable forever — Pro keeps new capture flowing.

Get Thread

Don't search for the chat.

Recover the thought.

Thread is built for people who use AI to think, create, research, decide, and build. The Mac app runs in your menu bar and works with the AI tools you already use.