AI news ·
Cyclon's AI phone aims to take over your apps without developer integrations
Cyclon is taking preorders for an AI phone built on a Google Pixel 8a, shipped within 30 days, that taps through apps for you. The open question is whether it handles every app's quirks.
Cyclon, a startup in Y Combinator's F26 batch, is taking preorders for a phone it says will replace your iPhone. The first batch is a Google Pixel 8a with Cyclon preinstalled, shipped within 30 days of preorder. The company's pitch: stop tapping through apps and let AI finish the task for you.
Most AI assistants need developers to build integrations before they can do anything inside an app. Cyclon doesn't wait for that. It moves through apps the way a person does and can jump between several to complete a job. Ask it to find photos buried in a WhatsApp chat and send them to a friend, and it will. When the app you need doesn't exist, Cyclon builds one on the spot.
Why it starts on a Pixel
Cyclon's work happens almost entirely at the operating system level, which is why the company launched on existing hardware instead of waiting for its own. The phone also keeps working in the background while it sits in your pocket.
Custom hardware is on the roadmap. With its own device, Cyclon plans to run more powerful AI models locally and pair them with cloud models, keeping as much personal data on the phone as possible. That's a data-privacy argument as much as a performance one.
The reliability question
Every app has its own layout and quirks, and an AI that taps through screens has to handle all of them. That's the open question for Cyclon and for anyone considering handing over their daily routines. If it works, it removes a lot of tedious tapping. If it doesn't, it becomes another assistant that promises more than it delivers.
For teams already testing AI agents inside their own tools, the same lesson applies: the demo is easy, the edge cases are not. Courses on AI agents and automation can help teams judge where that line falls before they commit budget.
Why this matters for your team
If an AI can operate apps the way a person does, the work that depends on switching between tools gets cheaper to automate. Customer support agents who copy details from a chat into a CRM, sales reps who pull photos or documents out of messaging threads, event planners who reconcile bookings across three systems: that's the kind of routine Cyclon is aimed at. The practical move now is to map which of your team's tasks are mostly app-switching and repetitive tapping. Those are the first candidates to test, whether on Cyclon or on whatever agent tool your company adopts next. The teams that know their own workflows in detail will be the ones who can tell a working agent from a convincing demo.