Introducing ThoughtBridge
Capture spoken thoughts anywhere, let AI file them where your tools can read them — and, when you're ready, turn a thought into a pull request.
Thoughts don’t respect contexts. The fix for the bug arrives in the car. The right opening line for the proposal shows up mid-run. The product idea lands while you’re making dinner, fully formed, and is gone by the time the pasta is drained.
It isn’t a discipline problem. Thinking happens everywhere; capture happens almost nowhere. And the few thoughts that do survive — a voice memo here, a note app there — end up somewhere none of your tools, and honestly little of your future attention, will ever look.
ThoughtBridge exists to close that gap.
Capture, then let it be filed
ThoughtBridge is a capture tool first. On iPhone, iPad, Apple Watch, or Mac, one tap starts a recording — voice or video — and you just talk. Offline is fine; captures queue and upload when they can. From the watch it’s a single tap from your wrist, no phone in reach.
Then the filing happens without you. Each capture is transcribed and analysed into a structured note: a title, a summary, and the tasks, questions, and decisions hiding inside it. Your exact words stay verbatim — the structure sits alongside them. Thoughts route to the right project automatically, and anything ambiguous waits in a review inbox rather than being filed by guess.
The last step is the bridge. Thoughts are written into destinations you control — a GitHub repo, Google Drive, or a local or iCloud folder — as plain Markdown, text, or JSON, with agent files so Claude, ChatGPT, and Cursor understand what they’re reading. The next time you open an AI tool in that project, your thinking is already on disk. One destination can host many projects, each in its own folder.
And then: “bridge it”
The part we’re most excited about. End a spoken thought with the words “bridge it”, and a self-hosted runner picks it up, drives Claude Code inside the project’s repository, and opens a pull request.
If the agent needs a decision, the question flows back to the app and you answer from your phone. The thought completes when the pull request merges. A thought spoken on a walk becomes reviewed, merged code — with you making exactly two decisions: saying it, and merging it.
The runner will be open source at launch. It runs on your machine, with your credentials, only on projects where you’ve enabled it. Nothing touches your code on our say-so.
Where it stands today
ThoughtBridge is in beta. The iPhone, iPad, Apple Watch, and Mac apps ship through TestFlight, and invites go out from the waitlist in batches. It’s free during the beta, with fair-use limits on captures.
The honest print: audio is deleted after transcription, thoughts live in the app and in destinations you own, and you can use it as a plain capture tool without connecting AI routing or destinations at all.
What’s next
A web app, for reviewing and promoting thoughts from any browser. The runner guide and its open-source release. More destination types as people ask for them.
If any of this sounds like the tool you’ve been improvising with voice memos, join the waitlist. We’d like to know what you’re thinking.