Revyl

Now I wonder how it'd look if you ask it to make a 3d structure with tubes of your app and all its parts Like that 3d SQL visualization from a few days ago Like seeing people visit your site, go through sign up and then seeing your robot workers send packages around, send welcome email, and all the other stuff Lots of fun ways to visualize your apps and operations in 3d

Google indexed the web. We index your app. Send as many agents as you want. They explore in parallel, and what comes back is a map of every screen they reached and every path between them. Now nobody has to guess how your app works or looks. Not your team, not your agents.

Everyone is wiring coding agents into Linear right now. Almost nobody is feeding those tickets real context. Give an agent "login sometimes fails" and it guesses. Give it the full trace, the failing step, and a screenshot of the exact moment, and it ships the fix. Revyl now creates the ticket from the error itself. Evidence travels with it. Context is everything. For humans and for agents.

If you build mobile apps and you've started handing work to Cursor's cloud agents, this one's for you. Here's the problem, those agents run on a Linux box in the cloud. Incredible for web. On mobile, they're flying blind. They can't run Xcode or a simulator, so they write iOS code they can never open, build, or see. The agent says "done" with zero idea if the screen even renders. So we gave your cloud agent a phone. It builds your app on a cloud Mac runner, opens it on a live iPhone, taps through the real flow, catches the crash, and comes back with a recording as proof. The loop finally closes on mobile, and none of it touches your machine. Why this matters, from experience: At Uber, a single p0 bug in production cost $2.5 million an hour. So we solved quality with people. Rooms of them, tapping through the app by hand before every release: add a card, cancel a trip, in 50+ languages, ship after ship. If I could have handed that to cloud agents that actually build and drive the app on live iPhones, we'd have caught those bugs overnight, in parallel, instead of staffing a floor to do it by hand. That's what this is. If you're building mobile with Cursor's cloud agents, it's the missing piece.

If you build mobile apps and you've started handing work to Cursor's cloud agents, this one's for you. Here's the problem, those agents run on a Linux box in the cloud. Incredible for web. On mobile, they're flying blind. They can't run Xcode or a simulator, so they write iOS code they can never open, build, or see. The agent says "done" with zero idea if the screen even renders. So we gave your cloud agent a phone. It builds your app on a cloud Mac runner, opens it on a live iPhone, taps through the real flow, catches the crash, and comes back with a recording as proof. The loop finally closes on mobile, and none of it touches your machine. Why this matters, from experience: At Uber, a single p0 bug in production cost $2.5 million an hour. So we solved quality with people. Rooms of them, tapping through the app by hand before every release: add a card, cancel a trip, in 50+ languages, ship after ship. If I could have handed that to cloud agents that actually build and drive the app on live iPhones, we'd have caught those bugs overnight, in parallel, instead of staffing a floor to do it by hand. That's what this is. If you're building mobile with Cursor's cloud agents, it's the missing piece.

Introducing Devin Outposts: run Devin on any machine. Your Mac mini, a GPU box in your lab, a VM inside your private network, or a Kubernetes cluster next to your internal services.

you've never actually seen your own app. you've seen one build. in english. logged in as you. on your phone. at Uber, that was the whole problem. no one knew what the app actually looked like. every week a room of designers would sit down and check, screen by screen, that the build matched the figma. and that was one language. the app shipped in 50+, across thousands of a/b tests, on hundreds of devices. thousands of versions of the same app, and nobody had seen most of them. we built atlas to map what users actually see.

At Uber, our test suite broke almost every week for reasons that had nothing to do with real bugs. Someone renamed a button. Someone moved a view in the hierarchy. The whole run went red while the app was completely fine. So the team stopped trusting it. Red just meant "probably the tests again," and that is exactly how real bugs slipped into prod. The selectors were the problem. So we deleted them. Now Revyl looks at the screen and taps what a human would. Rename anything, restructure the entire hierarchy, it still walks the flow. Red finally means something is actually broken.