Trust and security

Built in your name. Owned by you.

How we handle ownership, access and data. These are our own settings too, because we run the same way here.

System ownership

01

The system is yours from the first day.

The system, the agents, the instructions and the documentation are all yours. There is no licence to renew and no platform to stay signed up to.

If you stop working with us, nothing switches off.

You own everything on day one. Anything after launch is optional, and cancellable.

Accounts and infrastructure

02

Created in your name, on accounts you control.

The accounts are opened in your business name, with your billing. We do not build it on a Cainetic account and rent it back to you.

If you already have accounts you would rather use, we work in those.

  • Accounts are opened in your business name
  • Billing and admin sit with you
  • You can add or remove any user, including us
  • Nothing has to be migrated at handover
Client data handling

03

Your material stays in your accounts.

Your documents and records stay in your own storage. While we build, we work from a shared folder you control, so you can see everything we have and take it back at any time.

We keep a private backup of the work we deliver, the way any studio keeps its own records. It is never shared and never used for anything else.

Model-provider relationship

04

Built on Claude, made by Anthropic.

The system runs on Claude, made by Anthropic, on accounts you own. We are not the AI company and we do not resell one. Wherever possible the account is in your name, so your agreement is with Anthropic directly. Where it has to sit differently, we tell you first.

You are not tied to us to keep using it.

Training and data-use settings

05

Training on your material is switched off.

We set this deliberately, rather than leaving it on the default, and we show you where the setting lives in your own account so you can check it.

We never train anything on your material, and never use it for anyone else.

Access permissions

06

Permissions come before capability.

Before an agent can do anything, we set what it is allowed to reach. One job each, and a limit to match. Nothing gets wide access just because it was quicker.

  • Permissions are defined first, capability second
  • Each agent has a named responsibility
  • Each agent has an access boundary to match
  • Boundaries are written down, so you can check them
  • Access is reviewed whenever the system changes
Human approval

07

A person signs off anything that matters.

Anything that leaves the business needs a person to approve it. The agents prepare and check. Your people decide and send.

It is written into every agent, not a setting someone can switch off on a busy afternoon.

Our access, during and after

08

Our access is visible, and it can be removed.

We work as named users in your accounts, so you can see who has access and what they can reach. When the build finishes, our access comes off unless you ask us to keep it.

You remove us from your own account. You do not need our permission or our help.

The test we build to

If Cainetic disappeared tomorrow, the system would still be yours, documented, and working.

That is not a promise. It is how the thing is built.

Handover

09

Documented, then handed over.

You get the system and a document explaining it: what each agent does, what it can reach, and how to change it. Your people are trained on real work, not a demo.

Handover is the finish line, not a door into a retainer.

Ongoing Development access

10

Only as far as you scope it.

If you take Development, from $1,000 a month, we keep improving the system. We only get the access you give us, and you can stop at any time.

You choose this after the system is already yours and already working.

You own everything on day one. Anything after launch is optional, and cancellable.

What we do not do

11

The limits, stated plainly.

What we will not do matters more than what we claim.

  • We do not keep your live systems running on Cainetic accounts.
  • We do not train models on your material.
  • We do not use your material in work for anyone else.
  • We do not let agents decide or send anything on their own.
  • We do not claim security certifications. We hold none, and we will not pretend otherwise.
  • We do not name the businesses we have built this for without their permission.
  • We do not keep a key you cannot remove.
The next step

Ask us anything on this page.

If something here matters to you and it is not answered, ask. We would rather lose the job than overstate what we hold.

Or start where everyone else does, with the Map, and see what the system would take off your team first.

Ask us a question