Security
Your systems, handled with care.
How we handle your systems and your information while we work together, set out plainly.
Your live workflow stays in accounts you own
The working system is delivered into your own accounts and tools. We keep reusable build infrastructure, not client data, in our shared automation library. Any client-specific files or examples stay in a separate client workspace and are never reused for another business.
You own the AI account
You hold and pay for the Claude or OpenAI account selected for the work. We build inside that environment, so the delivered workflows and projects run on an account that belongs to you.
Access starts read-only and stays limited
We ask for the least access the agreed workflow needs. Your source systems stay read-only by default. If a workflow later needs to write something back, the exact action, scope and human approval gate are agreed before that access is enabled.
Builds are tested away from live work
We build and dry-run in a separate sandbox. A final test in your real environment happens only after the workflow passes the agreed checks, and the manual process stays in place until you accept the result.
An NDA before we collect your numbers
We sign an NDA before collecting your operational or financial figures. We only ask for the examples and access needed to map, test and verify the agreed workflow.
Nothing goes live without your sign-off
Every build has a written spec, and you sign acceptance against it before anything goes live. Irreversible actions keep a human approval gate.
Bring your questions.
We'll walk through any of these commitments point by point, with you or with the person who runs your systems.
AI Strategy Review