Miles Managed Cloud
We run, patch and scale the machines.
Deploy anywhere
Run AI agents on your own infrastructure
Each agent’s computer runs in Miles Managed Cloud, in your AWS, Azure or GCP account, or on your premises, air-gapped. Mix them per agent.
Book a call
FIG 5.0
Where agents’ computers run
Your cloud
Computers run in your AWS, Azure or GCP account. Only status, usage and what you watch leave it.
One switch
Stop every agent and sleep every computer with one typed phrase.
1.0Three places
Three ways to
deploy AI agents
Each option answers four questions on one screen: who runs it, where data lives, what leaves your network and who pays.
Miles Managed Cloud
We run, patch and scale the machines. Twenty hours per seat are in your plan.
Exceptions, with a reason
Scout browses untrusted sites, so it stays outside your network.
One agent at a time
Move an agent between places without touching the others.
Data residency stated
Each place says where screens, files and logs live.
Health per place
Running, provisioning or failed, with the next step.
A default place
New agents start there. Changing it moves nobody.
2.0Your cloud
Deploy AI agents in
AWS, Azure or GCP
A guided setup: pick the region, apply a CloudFormation or Terraform template, and test the connection. Your cloud bills the machines.
AWS, Azure, GCP
In any region your provider offers.
Template, applied unchanged
CloudFormation or Terraform with your IDs filled in.
Connectivity test
Miles starts and deletes a computer to prove it works.
Time stated up front
About 20 minutes, then provisioning you don’t need to watch.
What leaves, in writing
Status and usage, the live screen while someone watches, and what the agent sends to the model. Nothing else.
A boundary you control
Computers and their files stay in the region you pick.
Your data doesn’t move. The agents come to it: in your cloud account, or on your own hardware with no way out.
3.0Your premises
Air-gapped AI,
on your premises
Run agents on your own Macs and Linux hosts, or all of Miles inside your network with nothing going out. Each new machine waits for approval.
Revoke, and block it
Its key stops working, the record stays, and it can’t be added back until you say so.
Approval before work
A new machine gets no work until an owner or admin approves it.
One install command
With a token that expires in an hour.
Macs and Linux
Apple silicon and x86-64, as a system service.
Agent versions
Too old to take work is a state of its own.
$0 compute, stated
Hardware and power are yours; Miles charges its fee.
4.0Kill switch
Stop everything,
at once
One typed phrase stops every agent and sleeps every computer, wherever it runs. Nothing is deleted, and only an owner turns it off.
Stops at the step
Running tasks stop within seconds and keep their checkpoint.
Control ends
Anyone driving a computer loses control.
Requests declined
What was waiting for an OK can’t happen later.
On the record
Who, when, and who turned it off again.
Visible while it’s on
Who turned it on and when, and a way back that only owners have.
Or one fleet at a time
Sleep every computer without stopping agents.
5.0Who pays
Every bill,
said plainly
Managed cloud time comes from your plan’s hours. Your cloud bills its own instances. Your own machines cost nothing to compute.
Kept apart from Miles spend
What your cloud bills is shown, and left out of your AI budget.
Plan hours
20 hours per seat in Miles Managed Cloud, then by the hour.
Your cloud bill
Instances, storage and egress, estimated from awake time.
Self-hosted
$0 compute, tagged, never left blank.
One period
Every number uses the same billing period.
More of Miles
Open this week
Your data, your rules. Our agents.
Bring a real task. We’ll show an agent doing it on its own computer, in your apps, and asking before it acts.
+3
Book a call