ThreadPage
Start writing

All use cases

An incident, written down as it happens

Two lines whenever something changes, and each note keeps its minute. On Friday, @Claude reads the thread and drafts the postmortem’s timeline.

Start free trial

14 days of Pro. No card needed. Ends automatically. Choose to continue with Pro for $99.99 a year ($8.33 a month). See monthly pricing

Start writing
See plansSee the thread

At a glance

For
On-call engineers and whoever writes the postmortem.
You keep
A thread per incident in an Incidents folder shared with the team at Can add notes.
Comes in
  • Two lines from whoever is on it, at each step, with screenshots of the graphs.
  • A draft of the postmortem’s timeline from @Claude, on your own API key, when you ask.
The moment
On Friday @Claude reads the thread and drafts the timeline, from the first error at 21:40 to the error rate back to normal at 22:29. You start the postmortem from it.
Plan
Pro Only the one who shares needs Pro. The team joins free.

Before

Checkout errors start at 21:40. By midnight the story is spread over a chat thread of 140 messages and two video calls nobody took notes on. On Friday you sit down to write the postmortem, and the first question is when the rollback started.

Why a thread, not the incident chat

  1. The notes are the timeline

    Every note keeps its minute, so two lines at each step leave the timeline behind them.

  2. The next person starts from the last note

    At 02:00 you hand over. Ken reads the thread, not 140 messages.

  3. A draft instead of a blank page

    @Claude drafts the timeline from the notes in the thread. You check it and write why it happened.

A thread, over time Example

Your workspace starts empty.

Incidents

Checkout errors, 28 October

DoneSharedKen, Priya

  1. Wed 28 Oct

    1. The first note: what you see, with the graph.

      You21:43

      Checkout returns 502 for about 30% of orders since 21:40.

    2. Ken adds what he found, under his name.

      Ken21:52New

      Deploy 4512 went out at 21:36. It sets the database connection pool to 20; it was 100.

    3. You22:05

      Rolling back to 4511.

    4. Waiting while you watch, and the note says until when. Waiting

      You22:31

      Error rate back to normal at 22:29. Watching until 02:00, then I hand over to Ken.

  2. Thu 29 Oct

    1. The handover is the last note, so Ken starts there.

      You02:00

      Handing over to Ken: the rollback holds, and 4512 stays out. Open question: how the pool change passed review.

  3. Fri 30 Oct

    1. On Friday, one request on your own Anthropic key.

      You10:00

      To @ClaudeReplied in 52 sec

      @Claude draft the postmortem timeline from this thread, with the time of each step.

    2. Its answer is a note. You check it against the notes above.

      From @Claude10:01

      Draft timeline, from the notes in this thread:

      TimeWhat happened
      21:36Deploy 4512 goes out
      21:40Checkout errors start
      22:05Rollback to 4511 starts
      22:29Error rate normal

    3. Done, with the action that came out of it. Done

      You16:20

      Postmortem published. Action: a load test before any change to the pool size.

You ask. The answer is a note.

Each one reads and writes only where its access allows.

Connect an assistant

  • The timeline, drafted

    Write to @Claude in the incident’s thread. It reads the notes and answers with the timeline as a note, on your own API key.

    Set it up

    AI modelYour Anthropic or OpenAI API keyPro with Sync

    1. In Settings → Connections, open AI models, choose Claude or GPT, paste an API key from Anthropic or OpenAI and choose Connect. AI models need Pro with Sync turned on.
    2. In the thread, write @Claude and what you need, then choose Send. In a thread you share, choose Allow here the first time.

  • Alerts in a thread of their own

    If your alerting tool can send a webhook with a body you write, each alert becomes a note in an Alerts thread. The incident’s thread references the one that started it.

    Set it up

    AutomationAdd notes from another appText or Markdown

    1. In Settings → Connections → Automations, choose Connect an automation → Add notes from another app. Name it Alerts, choose the Alerts thread and Text or Markdown, and copy its URL.
    2. In your alerting tool, add a webhook to that URL that sends {"text": "…"} with the alert’s name, its state and a link to the graph.
    3. Send a test alert and read the note it adds.

Limits

  • ThreadPage doesn’t page anyone: alerts and calls stay with your on-call tool.
  • @Claude needs an API key from Anthropic, and the provider bills the key for every answer. In a thread you share, it answers only after you choose Allow here.
  • Only you can write to @Claude in a thread you share. The team reads its answers.
  • Write the numbers you read off a graph into the note: the model works from text more reliably than from a screenshot.
  • An endpoint URL feeds one thread and carries text, not images.
  • ThreadPage sends what the model reads to Anthropic with your key.
  • Notes aren’t end-to-end encrypted: keep secrets and customer data out of them.

Claude, GPT, Anthropic, and OpenAI are the names of their owners. ThreadPage is independent and not affiliated with or endorsed by them.

Start with one thread.

Start free trial

14 days of Pro. No card needed. Ends automatically. Choose to continue with Pro for $99.99 a year ($8.33 a month). See monthly pricing

Start writing
See plans

More use cases

  • Handing off before you leave

    Going on leave? @Claude reads the client’s folder and drafts a handoff with every promised date. You fix what it got wrong and share the folder with Nora.

    Pro @Claude and sharing

  • Release notes drafted from your pull requests

    Every merged pull request files itself in a thread. On release day @GPT drafts the notes from them, and the release you publish lands where customers read it.

    Pro GitHub and @GPT
All use cases