Services
Every engagement starts from the same question: where does your product's AI say more than it can safely stand behind — and what happens to a real user when it does?
That question can be answered at three depths. An audit diagnoses the risk. A design engagement builds the safeguards. A retainer keeps the judgment on hand as you go. All three are grounded in the real regulatory surface — HIPAA, the FTC Health Breach Notification Rule, and the FDA's software-as-medical-device lines — and all three produce something you can act on, not a deck.
Most teams start with an audit and go deeper from there.
Book a meeting
Audit
An audit maps every place your product's AI makes a claim a user could act on, and asks the question most teams haven't: how much of what it says can it actually stand behind? I classify each claim — fact, inference, or something the system can't really know — flag where the confidence outruns the evidence, and assess those surfaces against the real regulatory exposure: HIPAA, the FTC Health Breach Notification Rule, the FDA's software-as-medical-device lines, the state consumer-health-privacy patchwork.
You get a findings report, ranked by severity — what could genuinely hurt a user or expose you, and what to simply keep an eye on — with a clear direction for each. It's a diagnosis, not a redesign: you leave knowing exactly what needs attention and where, whether you take that to your own team or bring me back to design the fixes.
Responsibility Layer
The strongest moment to get your AI's judgment right is before any of it ships. Bring me the plan — even a rough one — and I map every claim your product will make on a user, establish what it can actually stand behind, and design the flows for the moments that carry the most weight: what the interface says when it's uncertain, and what it does when it's wrong.
You leave with buildable design your engineers can implement directly, and the reasoning behind each decision — so when your compliance review comes, the case is already made. Getting this right early means you ship on your own timeline, instead of retrofitting safety into a product that's already live.
Retainer
They come up mid-build — the feature you're deciding whether to ship this week, the claim you're not sure the model can stand behind, the edge case someone raised in standup. A retainer is having someone who already knows your product and its risk surface reachable when those decisions land: a substantive answer within one business day, and a standing call for the questions that need more room.
What you're paying for is judgment and direction — a read on your plan before you commit, not a postmortem after. When something needs to be designed or built, that's a separate engagement; the retainer keeps you advised, not staffed.
Because good ongoing judgment depends on knowing your product, the retainer begins after we've worked together — so the advice is informed from day one. Three-month minimum