Founder FAQ: AI in Regulated Healthcare Startups

Founders moving from research benches into health-AI startups hit the same questions over and over. Here are the ones we get most often, answered plainly.

Do I need FDA clearance before I can pilot?

Not always. Feasibility pilots in a single clinic, under a research protocol with ethics approval, usually don't require clearance. Commercial deployment is a different conversation — talk to a regulatory consultant before you sign your first paying customer.

Should the AI make the clinical decision?

No. The pattern that works is decision-support: the AI recommends, a named clinician decides. This is also what reviewers, insurers, and hospital procurement teams want to see.

Rule-based logic or LLMs?

Use rule-based logic for anything safety-critical — thresholds, escalations, dosing. Use LLMs for the soft layer: phrasing, patient-friendly explanations, summarization. Mixing them deliberately is a feature, not a compromise.

Who owns the IP if I came out of a university lab?

Usually the institution, with a licensing path to the founder. Read your tech-transfer office's standard terms before you incorporate, not after.

How do I find a clinical co-founder?

The credible profile is a practising clinician who has also shipped something — a registry, a tool, a protocol. Conferences, hospital innovation offices, and founder communities are better hunting grounds than LinkedIn cold outreach.

What does a good first pilot look like?

One site, one condition, ten to thirty patients, a clear primary endpoint, and a pre-registered analysis plan. If you can't write that on one page, the pilot isn't ready.

How do I fund the gap between research and revenue?

Non-dilutive grants first, then angel or seed once you have pilot data. Talking to other founders who've crossed this gap helps — that's a lot of what happens in The Hub and at Founders Pathway events.