the manifesto

Soon, everyone runs an AI employee — built to work, not to chat.

Models can already reason, plan and decide. What they can't do is reach where the work actually happens — inside the apps, behind the logins, on the phone. The missing piece isn't intelligence. It's a way in, and an identity to act through.

  1. 01

    First the computer. Then the browser. Now the phone.

    Computer-use agents proved models can drive a screen; browser agents took it to the web. But the economy runs one layer deeper — in apps, on phones — and that's where agents still can't go.

  2. 02

    In the AI age, identity is survival.

    An agent with no real device, number or accounts can't act — or be trusted. Machine identities already outnumber humans, and most are over-permissioned. We give every agent a real one: owned, auditable, in your control.

  3. 03

    Real phones — one, or a hundred thousand.

    Your own device, or cloud phones spun up by the thousand on demand — streamed, controlled and recovered. The body agents never had, at the scale a workforce needs.

  4. 04

    Real, not virtual.

    Emulators and virtual phones get detected and blocked. We run genuine devices on real networks — so apps see exactly what they expect, because it's real.

the stakes

A trillion-dollar shift — and agents are about to run it.

~94%
of mobile time is spent in apps, not the browser
15%+
of everyday work decisions made autonomously by 2028 — up from zero today
33%
of enterprise software will include agentic AI by 2028
~$60B+
influenced by AI agents in one Cyber Week — already
the stack

A mobile agent stack — agent-ready.

  1. control
    API, CLI, MCP and SDK — drive the phone from code, an agent framework, or plain language.
  2. orchestration
    A hosted orchestrator that plans and drives agents across the fleet — assigns, supervises, recovers.
  3. devices
    Real cloud phones, physical devices and your own fleet — streamed live, scaled as capacity.
  4. identity
    Real-device identity — numbers, eSIM and accounts an agent owns and acts through. Auditable.
  5. execution
    An on-device agent that reads the screen, acts and recovers until the task is done.
  6. app intelligence
    A model of how apps work, growing with every run — so agents arrive already knowing the terrain.
  7. reliability
    Persistent sessions, real networks, retries, recovery and human override.
  8. infrastructure
    Carrier networks, proxies, device lifecycle, uptime and observability.

The model decides. The Phone Layer executes.

who builds on it

Live today. Expanding into everything.

AI agents & platforms
Give any agent a mobile body — or embed the runtime inside your own platform.
Enterprise automation
Mobile-first RPA across the apps teams already use — no backend change.
Sales & outreach
Every rep agent on its own device, number and identity.
Social & content
Run the accounts you manage on real devices, end to end.
Commerce
Marketplace and retail operations across a device fleet.
Healthcare
Clinical notes into mobile EHR workflows — secure, human-in-the-loop.
Market intelligence
Competitor apps, pricing and event data — tracked on schedule.
QA & security
Test and pentest real journeys on real, isolated devices.
Gaming
A full-device runtime for game-playing agents.
the moat

Built on us. Not around us.

mobile identity
Real-device identity at scale — numbers, eSIM, accounts you own. The hardest to replicate, the easiest to trust.
hardware
A fleet of real Android & iOS, cloud and physical. Atoms, not API calls — you can't pip install a phone farm.
infrastructure
Carrier networks, credentials, lifecycle, uptime, observability — the plumbing nobody else wants to build.

A feature gets cloned. A dependency built on identity, hardware and infra doesn't.

Give it a decade and agents run on every surface — glasses, watches, every app on Earth. The layer that lets them act, reliably and with a real identity, becomes as fundamental as the cloud.

Two things no agent has: a body, and an identity. We build both.

Start free Get in touch