Skip to content

MOBILE

An agent in your pocket.

A native app for iOS and Android with a real agent running on the device itself. It signs in as you, shows up on the platform like any other agent, and asks for your fingerprint or face before touching anything sensitive.

A SESSION IN THE FIELD

Speak, approve, done.

Ask out loud. The agent plans on the device, and when a step needs the camera, your location or a write, it stops and asks you first. You approve on the device with biometric confirmation, and the result lands in the platform where your team can see it.

9:41
Field agent On device
Take a photo of the meter and log it
Plan
  • 1. capture the meter with the camera
  • 2. log the reading to the platform
camera_take_photo
confirmed on device
meter photo
OK
Logged to the platform · visible in your agent sidebar

On-device approval with biometric confirmation. Reads flow through; sensitive actions wait for you.

Explore Agents

A PEER, NOT A REMOTE CONTROL

The phone joins the platform like any other agent.

It signs in with your company sign-on and acts under your identity. Its sessions appear in the platform agent sidebar, its tools are listed where admins already manage every agent, and the same permission rules apply.

DEVICE TOOL PERMISSIONS
ADMIN · AGENTS
  • read screens & data allow
  • write / edit records ask first
  • camera biometric
  • location biometric
  • contacts biometric

Admins see and set the phone agent’s permissions in the same place as every server agent.

Your identity, everywhere
The phone acts as you. What it may touch is decided by the same roles that govern you on the platform.
Visible to your team
Its sessions and scheduled jobs show up on the platform, so mobile work is never invisible work.
Server tools, safely
When it needs a platform capability, it calls approved server-side tools through an authenticated bridge, with credentials kept server-side.

BUILT FOR THE FIELD

The tools a phone actually has.

The agent runs on the device itself, so it can use what the device can do: see, listen, locate, notify. Every one of those capabilities is a permissioned tool, not an open door.

Camera and voice

Take photos, listen to spoken requests and answer out loud. Field work without typing.

Location aware

Where-am-I context for tasks that depend on being on site, gated behind biometric approval.

Runs on the device

The agent loop executes natively on the phone, keeping sessions responsive on the move.

Your own Claude login

Sign in with your Claude account to power the agent. No API keys to manage.

Scheduled check-ins

Jobs can be scheduled to wake the app and run, so routine tasks happen without being asked.

Always current

The app updates over the air from your own infrastructure, so the fleet of devices stays on the same version.

CONTROL IN YOUR HAND

Nothing sensitive happens without you.

The mobile agent follows the same law as every T6X agent: reads flow through, writes wait for a person. On a phone, that person is you, and the confirmation is your fingerprint or your face.

Reads flow, writes ask
Looking things up is instant. Changing things, or touching the camera, your location or your contacts, stops and asks first.
A record that rides along
Tool use is tracked on the device in a tamper-evident record, so what the agent did is never a mystery.
Your account, your usage
The agent runs on your own model account. Usage is yours to see, and there is no markup in between.

TAKE IT WITH YOU

Your platform, wherever you stand.

Put a governed agent in the pocket of everyone in the field, with the same approvals and the same visibility your operation already runs on.