For studio and office managers · Out October 10
Never carry a client promise in your head again.
Octopus is a Mac app for the person who runs a small studio's office. It's built to remember what everyone promised, so you don't have to.
Sign-ups for drop day closed on Sunday 4 October, 12:00 pm PDT.
Drops Saturday 10 October, 10:00 am PDT.
The waitlist
Join, and get two things
- Right now
The Nothing-Slips Checklist
One printable page: a 5-minute end-of-day sweep to catch promises before they slip. Use it today, no app needed.
- October 10
The test build
Everyone who signs up by October 4 gets the test build on October 10. Mac with Apple silicon, macOS 14 or later.
What it does
- One list. It reads what you and your clients promise in Mail, Slack, Messages and your notes, and keeps one list of what you owe people and what's coming.
- The dates on your calendar. Each one goes on as a hold that invites nobody.
- Later: follow-ups ready each morning. Overnight it prepares them. Before one goes out, it tells you what it will send and when, with a Stop button. It's in testing from mid-November, with offices already using the list.
The honest part
- Out Saturday, October 10. Everyone who signs up by Sunday, October 4 gets it on the 10th. It's free on your own Claude or ChatGPT plan. Installs on your Mac in about 15 minutes.
- It will catch a lot, not everything. When it misses something, tell me and I'll fix it.
- Octopus runs on the AI you already pay for. It needs Claude Code or Codex on your Mac.
- It does the work, and tells you first. Before it sends, pays or deletes anything, or invites anyone, it says what and when, with a Stop button. If you do nothing, it goes ahead. If a date, amount or other number isn't in your own messages, or a value isn't on your screen, it asks instead of guessing. Want to approve every step? Turn that on.
- It needs a Mac. One you use every day. It doesn't run on Windows or on a phone.
- It reads your screen. To find promises it reads the text of what you're working on, and sends short pieces to an AI model online to make sense of them.