Clinical trial enrollment is slow, with some protocols taking up to eight months to find a patient. Clinicians spend a lot of time on administrative tasks. Even though medical AI has seen significant investment, most tools operate as “black boxes,” producing results without clear explanations, which limits their use when accountability matters.
Health Universe, a San Francisco startup founded by Dan Caron, just closed a $6 million seed round led by Kleiner Perkins, raising total funding to $9.5 million. The company’s core claim is straightforward: it builds healthcare AI agents that maintain a clear audit trail.
Will that promise stand up under real-world pressure? The $6 million investment alone doesn’t prove it yet.
What “auditable AI” actually means, and why it’s harder than it sounds
In regulated industries, auditability means more than just getting the right answers. Regulators like the FDA, IRB, or ONC need to see the full reasoning process: the data used, the model version, and any human checks before approval. Most commercial AI tools focus more on usefulness than on being transparent.
Health Universe holds ONC B.11 certification under the 21st Century Cures Act for decision support interventions. B.11 mandates transparent, clinician-reviewable logic and allows users to opt out of recommendations. This differs from HIPAA attestation and does not equate to FDA clearance for diagnostic clinical decisions.
Health Universe is tackling this with ‘source-linked’ outputs: every summary or recommendation points straight back to the original clinical records or data. They’ve also built Observer, a dashboard that tracks errors and query costs and flags risky outputs. Not something you see every day in enterprise AI.
The full stack spans Navigator (a secure workspace for deploying agents via TEFCA or FHIR), Explorer (for building population-level cohorts), and custom agents for oncology chart review and drafting clinical trial protocols.
In a pilot with the Duke Clinical Research Institute, Health Universe slashed clinical trial setup time by 30-40 times, saving 93% of the total time. If those numbers stick, we’re looking at one of the biggest leaps in clinical research efficiency in years.
The direct competitors include Nabla, Tempus, and Insilico Medicine.
Why Kleiner Perkins led and what it signals
Kleiner Perkins has a track record with digital health bets like One Medical and Devoted Health. By jumping in at seed instead of waiting for Series A, they’re making a statement: the FDA is watching AI in healthcare more closely, ONC’s B.11 framework is only going to get bigger, and the regulatory bar is rising.
So, Health Universe is betting on the compliance infrastructure every AI company in this space will need sooner or later.
What comes next — and what’s still unproven
The $6M will help Health Universe expand into more academic medical centres and build out an Agent-to-Agent (A2A) Marketplace, a network where AI agents from different organisations can share documents, but with controls in place.
If it works, this could be the start of a federated AI backbone for healthcare. For now, though, it’s still more vision than reality.
