AI, on the data you are not allowed to be careless with.

Organizations want to use AI on data they cannot be careless with — patient records, client files, anything a procurement questionnaire asks about. We work out what your obligations actually require, then build one workflow against those controls so the answer is evidence rather than advice.

• Two weeks, CA$9,000 fixed, ending in a written position and one workflow running.
• Then build the rest to the same standard, at a fixed price, or take it in-house.
• For clinics, law firms, suppliers to BC public bodies, and anyone holding a security questionnaire.

An agent usually shows up holding one person's entire access. Point it here instead and it inherits exactly what that person's groups allow — no shared keys, no standing credentials, nothing new for you to administer. Take someone out of a group and their agents lose that access with them.

• A tool the caller may not use is simply not offered, so nothing can talk the agent into reaching for it.
• Anything destructive stops and waits for a named person to approve it, without blocking everything else.
• Every action leaves a signed record that shows whether anyone has edited it.

The engagements above are delivered on our own platform: the records, the rules, the screens and the approvals are generated from how a business actually works, so a change is a change to the rules rather than a rebuild. The pieces that are useful on their own are open source and permissively licensed.

• SDKs in Python and TypeScript for teams building their own agents.
• The audit and receipts layer, MIT and Apache-2.0 — take it, ship it, never talk to us.
• Not sold as a product yet. Today it is how the work gets done.