Deployment and integration

We let agents operate machines that have no API and cannot run agent software.

An agent that works a machine the way a person does, with nothing installed on it. Making that hold up on real hardware is the job: choosing the approach, preparing the equipment, and proving the result is quick and reliable enough to leave running. That is what we do.

High Performance General Purpose Agent Machine Interface

Talk to us about a deployment What an engagement covers

The complete picture in one screen.

The target sees a monitor and a keyboard and cannot tell the difference, so every screen a person can operate becomes operable, including the ones that exist because no software is running yet.

The full range of what a keyboard and mouse can do, not a crude approximation of it.

One way in, and it is locked. No quiet exception for anyone already on the machine.

Faster than a person doing the same work, and steady enough to leave running. Figures and caveats under What we measure.

The loop

What the target sees

The machine cannot tell it is not a person.

Every screen a person can work, the agent can work. Nothing is installed on the machine, nothing is granted to it, and nothing is asked of it that someone sitting at its keyboard would not ask. It answers the agent exactly as it answers anyone who sits down in front of it.

Nothing has to be prepared first, and nothing has to be true about the machine beforehand.

Locked, unprovisioned, mid-install or long out of support, it makes no difference.

Whatever the agent does is something a person could have done at the same keyboard.

Where an API stopsEveryday jobs agents cannot reach, because the machine has no API.ComparedHow frontier-lab computer use differs from AMI, in seven points.ExamplesThe eight shapes of job this reaches, and where it is the wrong tool.What to expectWhat it does well, what it cannot do, and what surprises people.Agent costsWhat an agent actually costs to run, and how the design controls it.SecurityThe threat model, stated first, and where the boundary really sits.What we doThe six steps of an engagement, from assessment to handover.What we measureFrame rate and action-to-visible, and what the numbers do not prove.Running it yourselfBuy the parts instead, and what that takes on.