Launching today
Kin
The AI labs want to own you. Make them answer to you.
50 followers
The AI labs want to own you. Make them answer to you.
50 followers
Let your AIs run free. Kin keeps them in check. Connect GitHub, Gmail or Stripe once, add Kin to ChatGPT, Claude or Cursor, and give each action a rule: Allow, Ask me or Block. Ask me waits for your OK on the web or your phone.
Hey everyone, I'm Troy, and I built Kin.
So here's why. I use a bunch of different AIs every day and I got really frustrated. I was constantly context switching between them, there was no coordination, every app handled approvals differently, and I was worried about things going out under my name without me knowing.
So I built one place for all of it, and that's Kin. It sits between my AIs and my accounts. I connect GitHub, Gmail and Stripe once, add Kin to every AI app, and give each action a rule: Allow, Ask me or Block.
I use it daily for my own work. All my AIs go through it, they can pick up each other's work, and when a new AI comes out it gets my accounts under the rules I already set. Nothing goes through Kin under my name unless my rules allow it or I approve it from my phone. And my AIs never see my passwords or keys. Kin keeps them encrypted.
And it's bigger than approvals. I can let my AIs run free and know Kin keeps them in check, and no single lab owns my setup anymore.
It works with ChatGPT, Claude, Codex, Cursor, Gemini and Grok. Pro is free for 7 days with no card in the US and 33 other countries, and Free works most places, just not in Europe yet. It all works on the web, there's a Kin app for the Mac and a bridge for Windows and Linux if you want your AIs on your computer too, Android is on Google Play, and the iPhone and Watch apps are on TestFlight.
I'm really excited to get this in front of people, so let me know what you think and what you'd want it to do next.
hope today brings you good first users. my agents run on auto nearly all the time, so Block matters most to me. does the agent get told why it was blocked?
@olia_nemirovski Thank you, I hope so too! And yes, it does. When something's set to Block, the agent gets a clear answer back instead of a silent failure. It's told you blocked that action, that it won't run, and not to keep retrying, so it can tell you or find another way to get the task done. That way an agent on auto doesn't just loop on it, and the attempt still shows up in your activity log.
Block's only part of the equation though! You can also set things to Ask me, so the agent waits for your OK on your phone, which is nice for the stuff you don't want fully off but don't want running on its own either :)
Have some more features coming soon like time-delayed allow and allow but notify. Because Kin allows undo on certain actions taken, it can let you auto approve some things with the peace of mind you can undo them or have some time to block something!
Looks cool. Do I understand correctly that it's kind of like a layer of fine grained tokens on GitHub but for any app?
@pronskiy Yeah, that's part of it! It's like fine-grained tokens but per action, and for every account you connect, not just GitHub. Your AIs never hold your GitHub token, or any other account's. They only sign in to Kin, and Kin keeps those tokens and checks your rules before anything runs.
But that's one piece of my vision and ambition here. The bigger thing for me is having one place all my AIs work through. They can pick up each other's work, a new AI gets my accounts under the rules I already set, and I can see everything they did and pause them all at once if something looks off.
Also have some pretty exciting stuff planned for enterprises to tackle defend against agent swarm cybersecurity attacks!