What we commit to

• Your data is never used to train shared models. Anything we build or tune on your operations serves you and only you. Nothing learned from your business leaks into anyone else’s.
• The intelligence stays yours. The systems we ship are built on your data and your workflows, and what they learn belongs to you, not to a vendor and not to us.
• We build in your environment. Wherever possible, production systems run in your cloud accounts under agreements you control, so your data stays where your policies already govern it.

How we work during an engagement

Every engagement runs under NDA before we see anything. From there, the rules are simple:

• Least privilege. Named individuals get the narrowest access needed for the function we’re working on. You can revoke it at any time, and it ends when the engagement ends.
• Data minimization. We work with what the system needs to be built and evaluated, nothing more. We don’t warehouse copies of your business.
• Encryption everywhere. Anything we handle is encrypted in transit and at rest.

Built on frontier AI, under enterprise terms

We build on the best frontier AI for your work: Anthropic, OpenAI, Google, or Microsoft, through their enterprise and API platforms. Under those terms, your prompts and outputs are not used to train their shared models. We deploy on the lab your team already trusts, inside agreements your security team can review before any data moves.

For high-volume, sensitive, or latency-critical work, we help you build your own intelligence and run it: models tuned on your data, serving only you. The intelligence stays yours.

Report a security issue

If you believe you’ve found a vulnerability on this site or in anything we run, email us. It goes straight to the founder and gets read the same day.