Launched this week
Glasshand
Give your agent a real iPhone and Mac
683 followers
Give your agent a real iPhone and Mac
683 followers
Glasshand gives your AI agents real iPhones and Macs to work on. Connect Claude Code, Codex, or any other agent, and it can rent a device with a single prompt. The hardware sits in our data center, and your agent uses it the way you would: posting on TikTok, booking an Uber, or building and testing iOS apps. Each device is yours alone for as long as you rent it. When you’re done, we erase your data and factory-reset it.
Free Options
Launch Team / Built With
CloudflareStill paying to deploy your apps? Try Cloudflare Workers for free
Promoted
This is exactly what my AI employees need. Incredibly good price.
@pavel_hegler Yes! We want to make sure it's affordable to everyone to try!
Congrats! And I guess your glasshand is agile and featherlight, isn’t it? Haha
@jojo_li and transparent
Cool. I'll try this. Congrats on this launch!
Real hardware makes a lot of sense — the simulator falls apart exactly where an agent would need it most, around system permissions and real device state. How far does that go for apps that need HealthKit, location or notification permissions, or a paired Apple Watch? Can the agent work through those prompts and test a flow that depends on them, or are Watch-dependent flows still out of reach?
Interesting angle: real devices instead of emulators. As someone who juggles several Apple IDs and Google accounts across companies, I'm curious how you handle sign-in friction on a rented device, e.g. 2FA prompts to the owner's phone and Activation Lock/Find My when the device is reset between renters. Is that part of the reset flow you described?
@modiyilreji This is exactly like setting up a new phone for yourself, you can go through the two of eight prompts and respond them on your own phone to finish the setting up process.
Real devices make sense for apps that fingerprint emulators, but I wonder about the other side. TikTok and Instagram are pretty aggressive about flagging automated behavior, so does an agent tapping around from a data center IP on a rented SE put the user's account at risk? Curious if you have seen bans or step-up verification during testing.
@galdayan In our testing so far, we haven’t seen TikTok or Instagram ban an account when the agent does the kinds of things you’d normally do by hand.
How you use the phone matters too. Sending the same DM to a hundred people can get you flagged even if you tap Send yourself. And if you’d prefer to use a carrier connection, you can put an eSIM on the phone and use mobile data.
Giving agents real devices instead of another API is an interesting shift. Curious how you handle those unexpected moments when an app asks for Face ID, a verification code, or human confirmation?
@jamieg You can set your own password that only you know. So if any app asks for a password or face ID, you can do an alternative of inputting the password. If it needs verification code or human confirmation, either the agent can do it or you can just operate on the phone. You also have the full control of the phone screen, and you can click around.