Two founders are building software for emergency department triage. They are excellent engineers. They have raised a seed round. They have interviewed twelve emergency physicians and built a product those physicians said they wanted.
The product assumes a workflow that does not exist.
Not a slightly different workflow. A workflow that exists in the descriptions clinicians give when asked in an interview, and not in what actually happens at 11pm on a Saturday when the department is at 140 percent capacity, the tracking board is being used in a way nobody designed, and the charge nurse is running a parallel system on a whiteboard because the official one is too slow.
The founders could not have known this. Nobody told them, because the physicians they interviewed described the workflow as it is supposed to be, which is what people do when asked to describe their work in the abstract.
What would have prevented this is not more interviews. It is one clinician on the team who owns that workflow and would have said, in week two, "that is not how this actually works."
The market has already reached this conclusion
The interesting development is that this is no longer a contrarian view. The capital market has priced it in.
Industry reporting on 2026 digital health funding describes "deep domain expertise and clinical background alignment" as having "become mandatory to clear venture capital due diligence."
That is a substantial shift. A decade ago, a strong technical team with a plausible healthcare thesis could raise. Now the absence of credible clinical depth on the founding team is a diligence failure.
The reasons are visible in the outcomes.
A JMIR analysis of 224 venture-backed digital health companies that had collectively raised $8.2 billion found:
- 44 percent had a clinical robustness score of zero.
- And robustness was essentially uncorrelated with funding raised, with an r-squared of 0.08.
Nearly half of a funded sector with no clinical evidence at all, and the amount raised told you nothing about whether a company had any.
And the most instructive failures were not scientific failures. Two prominent digital therapeutics companies, each with FDA-cleared products and each valued around $1.5 billion, went bankrupt. Not because the science failed. Because adoption did.
The products worked and nobody used them, which is a workflow and organizational problem, which is precisely the domain a clinical founder holds and an engineering team does not.
Why software cost collapsing made this worse
There is an economic reason this has become acute now, and it is worth being explicit about.
Building software has become dramatically cheaper and faster. What required a team and a year now requires fewer people and less time. Machine assistance has accelerated this further.
Which means the technical build is no longer the scarce input. When everyone can build, the differentiating asset is knowing what to build, and in healthcare that knowledge is almost entirely tacit, held by people doing the work, and unavailable from any document.
Industry commentary reflects exactly this shift, noting that vendors now compete on founder expertise and workflow integration rather than on features.
Domain truth is the scarce input. And the market has no mechanism for locating it.
The verification vacuum
Here is where it becomes a coordination problem rather than a talent problem.
Suppose you accept that a founding team needs genuine clinical depth. How does anyone verify it?
Doximity verifies clinical identity at scale and is not a founder marketplace. It can tell you someone is a physician. It cannot tell you they want to build something.
LinkedIn is a founder marketplace with no clinical verification. Anyone can describe themselves as a physician innovator, and the title carries no information about whether they currently practise, in what setting, or whether they have ever used the workflow in question.
Accelerators verify neither, in general.
So both sides are operating without verification:
The technical founder cannot confirm that their prospective clinical co-founder actually practises in the relevant setting, currently, or has any credibility with the buyers they will need to reach.
The clinician cannot assess whether their prospective technical co-founder has ever shipped anything.
And the investor running diligence on "clinical background alignment" is largely evaluating a CV and a conversation.
The predictable result is the arrangement everyone in this sector recognizes: a chief medical officer title granted for a quarter of a percent of equity, in exchange for roughly two hours a month. That is not clinical depth. It is the appearance of it, and it satisfies a diligence checkbox without solving the problem the checkbox exists for.
The other side of the same wall
The failure runs in both directions, and the clinician side is less discussed.
There are practising clinicians with genuinely valuable product insight, derived from years inside a specific workflow, who cannot get anywhere near a founding team.
Their obstacles:
They cannot find a technical co-founder. Their professional network is entirely clinical. A physician in a community hospital in the Midwest has approximately zero connections to senior engineers, and cold outreach from an unknown physician reads as noise.
They cannot assess technical credibility. Presented with an enthusiastic engineer, a clinician has no basis to judge whether this person can build the thing.
They are told to leave medicine. The standard advice to a clinician with an idea is to step back from practice, which is exactly wrong, because their value is the currency of their practice knowledge. A clinician who stops practising begins depreciating as a domain asset immediately.
And they lack the surrounding expertise. Commentary from physician-founders consistently identifies the same gaps: regulatory pathways, reimbursement strategy, fundraising, and operational support. None of which is taught anywhere in medical training and all of which is assumed knowledge in the startup world.
Research on physician entrepreneurship in the NHS context found ventures that "originated as intrapreneurial initiatives but shifted to individual-level resourcing", meaning institutional support evaporated and the clinician was left funding and organizing it personally.
So the domain knowledge that the market now says is mandatory sits with people who have no route to the market.
Proximity is doing the allocation
The mechanism connecting these two populations today is personal proximity, and it produces exactly the distribution you would expect.
A founder in San Francisco or Boston can find a clinical co-founder through their existing network, which will draw from a small number of academic medical centers.
Community clinicians, international medical graduates, physicians outside major startup hubs, and nurses and pharmacists with deep operational knowledge are effectively excluded, not by any judgment about their expertise but by never being in the room.
And there is a specific irony in this. The health systems where most American healthcare actually happens are community hospitals and non-academic settings. The clinicians most likely to know what will work in the average hospital are the least likely to be recruited to build it.
What would actually work
A verified vouching edge, not a title. The signal that matters is not "chief medical officer" on a slide. It is: this person practises in this setting, currently, and ten colleagues who work alongside them confirm they understand this workflow deeply.
That is a checkable claim. Nothing currently checks it, and it would be more informative to an investor than any credential.
Use the co-training graph. A clinician who cannot code but who can be vouched for by their residency cohort is a far stronger signal than a self-described physician innovator with an impressive profile. The referral literature establishes that co-training produces a durable, behavior-changing trust edge, which makes it a real verification substrate rather than a social one.
Match on workflow, not on specialty. "I need an emergency physician who has personally used triage software in a 400-bed community hospital" is a specific, findable requirement. "We need a clinical advisor" is not, and produces the two-hours-a-month arrangement.
Keep the clinician practising. Their entire value is currency of practice knowledge. Any arrangement requiring them to stop is destroying the asset it was formed to acquire. Part-time, protected, and ongoing is better for everyone.
And distinguish the roles honestly. A ceremonial advisor, a working advisor, and a co-founder are three different things with three different compensation structures, and conflating them is how both sides end up disappointed.
What you can do now
If you are building in healthcare without a clinician
Recruit for the specific workflow, not the specialty. The physician you need is the one who does this particular thing, in a setting like your target customer, right now.
Prefer community settings for product truth. Academic centers have different staffing, different resources, and different tolerances. If your customer is a 250-bed community hospital, an academic advisor may actively mislead you.
Watch the workflow rather than asking about it. The gap between how clinicians describe their work and how they perform it is the single largest source of product failure in this sector. Interviews capture the description. Observation captures the work.
Structure the role honestly. If you need dozens of hours, say so and compensate accordingly. A quarter of a percent for two hours a month buys a name, and you will find out what that is worth during your first health system evaluation.
If you are a clinician with an idea
Do not stop practising. Your practice is the asset. Reduce hours if necessary; do not exit.
Get vouched, specifically. Ask colleagues who have worked alongside you to say plainly what you know and have done. That is more persuasive to an investor than any self-description, and it is the thing nobody currently collects.
Find the technical partner through people who can assess technical work. Your clinical network cannot evaluate an engineer. Someone in your extended network can, and asking them to assess a candidate is a small, specific favor.
Learn the three things that are always missing: regulatory pathway, reimbursement mechanism, and who actually signs the purchase order. Physician-founder commentary identifies these consistently and none of them are intuitive.
Be honest about what you are. A clinician who deeply understands one workflow is enormously valuable and is not automatically a chief executive. Both roles are respectable and conflating them has ended a lot of promising ventures.
If you invest
Verify practice, not titles. Does this clinician currently practise, in what setting, and how much? A quarterly-hours clinical adviser with an academic affiliation is a different asset from a full-time community physician who owns the workflow.
Ask who vouches for them and talk to those people. This is more informative than any interview with the clinician themselves.
Ask what the clinician has actually contributed. Specification decisions, workflow corrections, introductions that converted. A clinical founder who cannot point to specific product decisions they shaped is ornamental.
Frequently asked questions
How much clinical evidence do digital health companies have? Less than funding levels imply. A JMIR analysis of 224 venture-backed digital health companies that had raised $8.2 billion found 44 percent had a clinical robustness score of zero, and that robustness was essentially uncorrelated with the amount raised, with an r-squared of 0.08.
Is clinical expertise now required to raise funding in digital health? Industry reporting on 2026 funding describes deep domain expertise and clinical background alignment as having become mandatory to clear venture capital due diligence, reflecting a significant shift from earlier periods when a strong technical team and a healthcare thesis were sufficient.
Why do digital health companies with approved products still fail? Frequently because of adoption rather than science. Two prominent digital therapeutics companies, each with FDA-cleared products and valuations around $1.5 billion, went bankrupt, with failure attributed to adoption and business model rather than to the underlying evidence.
How do healthcare startups usually find clinical co-founders? Through personal proximity: existing networks, accelerator communities, and introductions from investors' portfolio physicians. This systematically favors clinicians at a small number of academic centers in major startup hubs and excludes community clinicians, international medical graduates, and non-physician professionals with deep operational knowledge.
Can you verify that a clinical co-founder actually has relevant expertise? Not through existing tools. Clinical networks verify identity but are not founder marketplaces, professional networking platforms have no clinical verification, and accelerators typically verify neither. Peer attestation from colleagues who have worked alongside the clinician in the relevant setting is a checkable signal that nobody currently collects.
Should a clinician leave practice to build a company? The common advice is to step back, and it is frequently counterproductive, because the clinician's value to the venture is the currency of their practice knowledge, which begins depreciating immediately on leaving. Reduced but ongoing practice generally preserves the asset the venture was formed to use.
The bottom line
The healthcare technology market has concluded that clinical depth on a founding team is mandatory. It reached that conclusion by watching nearly half a funded sector operate with no clinical evidence, and by watching companies with cleared products and billion-dollar valuations fail on adoption rather than science.
Having reached that conclusion, it has no way to act on it.
There is no mechanism to verify that a clinical co-founder currently practises in the relevant setting, no mechanism to check whether their colleagues would vouch for their workflow expertise, and no mechanism for a community physician with genuinely valuable insight to reach a technical team that needs exactly what they know.
So the market settles for a title on a slide, granted for a fraction of a percent, in exchange for a couple of hours a month.
Meanwhile the clinician who could have said "that is not how this actually works" in week two, and saved eighteen months and a seed round, is working a shift in a community hospital four states away, has never met a founder, and has been told that if she is serious about her idea she should consider leaving medicine.
Part of a series on the missing professional infrastructure of healthcare. Previously: Pilot Purgatory
Evidence note: clinical robustness figures come from JMIR (2022) covering 224 venture-backed digital health companies. Funding environment and diligence expectations come from Rock Health reporting as summarized in trade press (2026). Company failure examples are drawn from a JMIR systematic review of growth-stage digital health success factors (2024). Physician-founder resource gaps come from published physician commentary and from Social Science and Medicine (2023) research on NHS clinician entrepreneurship. Co-training trust effects come from Pany and McWilliams in Health Services Research (2021).
