Until

we support this 𝕏 post

hehe. us too :)

AI agents now start about 4 in 10 pull requests on GitHub. Writing code got fast. Checking it didn't. The hard question in review isn't "is this good code?" It's "is this what we actually asked for?" Until.dev/x

they always come back 🥹 until.dev/apply/x

they always come back. Until early access → until.dev/apply/x

Your mom called. until.dev/apply/x

Your agent can write the code. It shouldn’t be making all the decisions. Until is where you work out the plan, the trade-offs and the review before the agent starts. Then it builds against what you decided. Start engineering again. DM us “Until” for early access. PS... We’re onboarding a small group first. Limited seats!!!

Less babysitting. More engineering. Join the line for early access → until.dev/x

Coding agents are getting ridiculously good. But keeping track of what they're actually doing is becoming a job of its own. That's a big part of why we're building Until. It's a ticketing system built for developers who work with coding agents. You decide what needs to get done, give your agents the jobs, and let them get to work. If something needs your input, they ask. They don't just make the call for you. The whole point is pretty simple. You should be able to hand off the work without handing over the decisions. You decide what to build. AI writes the code. We're opening early access to 500 developers, and the list is already filling up. If you're using Claude Code, Codex, or Cursor, DM us. We'd love to have you try it.

AI should be writing your code, not deciding what you're building. that's the idea behind what we're building at Until. you do the thinking. AI does the doing. early access is opening soon if you wanna check it out: until.dev/x

the future of building with agents starts with a decision. meet until. a git-backed ticketing system for people and coding agents. plans, jobs, reviews, and decisions, all in your repo. your agents do the jobs. you make the calls. until.dev/x