The cold email that goes nowhere
You’ve built a working clinical AI product. The validation looks good. You have a pilot site lined up. You’re ready to find your first ten customers.
So you do the obvious thing: you build a list of medical directors at midsize hospitals, you write a tight cold email, and you start sending. You get a 1.2% reply rate. None of the replies turn into meetings. None of the meetings turn into pilots. After a quarter, you have spent meaningful runway on a channel that produced zero customers and one suspicion that this whole thing might be harder than you thought.
It is harder. Not because clinicians are uniquely difficult buyers — they’re not — but because the things that make a clinical AI product worth buying are the same things that make it impossible to sell through a normal funnel.
This is the third post in our series on what makes medical AI startups structurally different. Post one covered the work. Post two covered the team. This one is about the customer.
Why normal funnels fail in clinical AI
A normal SaaS funnel assumes three things: the buyer can evaluate the product in fifteen minutes, the cost of trying is low, and the cost of being wrong is recoverable. Strip those assumptions away and the funnel collapses.
In clinical AI, none of them hold:
- Evaluation takes weeks, not minutes. The buyer needs to see your validation data, your model card, your security review, your DPA, your IRB letter if applicable, your evidence of subgroup performance, your integration plan. A demo doesn’t compress this — it just gates it.
- The cost of trying is enormous. A real pilot involves IT integration, a data-use agreement, clinician training, workflow changes, and a risk review. Hospitals don’t “try” things casually. By the time they’ve agreed to a pilot, they’ve already half-committed.
- The cost of being wrong is reputational and clinical, not just financial. A failed pilot doesn’t just waste budget — it teaches clinicians not to trust the next AI tool that walks in the door. Buyers know this. They behave accordingly.
So the normal funnel — awareness → demo → trial → close — doesn’t work because there is no “trial” stage that costs less than a full pilot. And there is no “demo” stage that gives the buyer enough confidence to commit to a pilot.
What you need instead is a longer, slower, more expensive top-of-funnel that does something the normal funnel never has to do: change how the buyer thinks.
The asymmetry: clinicians buy from people who teach them
Here is the thing the cold-email playbook misses: clinicians, more than almost any other professional class, learn from each other constantly. CME, grand rounds, journal clubs, conference talks, case discussions, society newsletters — clinical practice is shaped by a continuous stream of peer education.
This means there is a channel that already exists, that already has the buyer’s attention, that already has the right cultural posture (educator-to-peer rather than vendor-to-customer), and that almost no clinical AI startup uses well.
That channel is education. Not “thought leadership” — actual, practical, useful education. The kind that a clinician finishes and feels smarter, not pitched.
The startups that win clinical AI customers reliably almost all have the same shape underneath: the founder, or someone on the team, has been quietly teaching the buyer for six to eighteen months before a sale ever happens. Through a newsletter. Through a free email course. Through workshops at society meetings. Through case write-ups on a blog the buyer’s chief resident sends around. Through a podcast that another clinician told them to listen to on the drive home.
By the time the conversation about purchasing happens, the buyer already knows you. They already trust your judgment. They already think of you as a peer who has thought hard about this problem, not as a vendor with a deck.
What the curriculum looks like
The most underrated investment a clinical AI startup can make in its first year is a curriculum. Specifically:
A free email course — five to seven emails, one per day or one per week, that teach the buyer something concrete about the clinical problem your product solves. Not about your product. About the problem. How to think about it, what the evidence says, where the common mistakes are, what to ask a vendor. By the end of the course, the reader should feel like a more capable clinician, regardless of whether they buy anything.
A newsletter — a regular, high-signal dispatch on the intersection of clinical practice and AI. Not weekly if you can’t sustain it. Monthly is fine. The bar is not frequency, it is usefulness. Every issue should teach something the reader couldn’t have gotten from a conference abstract or a vendor blog.
Case write-ups — short, anonymized stories from real clinical AI deployments (yours and others’). What worked, what broke, what surprised everyone. These get forwarded. Forwarding is the entire game.
A workshop or talk circuit — show up at the small conferences your buyer attends. Not the big AI ones — the small clinical ones. Speak about the problem, not the product. Take questions. Answer them honestly. Leave behind a curriculum the audience can take home.
A code-shareable artifact — a reference architecture, a checklist, an open-source component, an anonymized dataset description. Something a technically curious clinician can hand to their hospital’s data team and say “this is what I’m thinking about.”
This is not a “content strategy.” Content strategy is what you do when you are competing for keywords. This is a curriculum, and it is what you do when you are building professional standing in a community that buys from people it respects.
The economics
Founders look at this and balk because it sounds slow and expensive. It is slower than a cold-email funnel. It is not expensive — it requires time and rigor, not budget — and it has an economic property that no paid channel ever has: it compounds.
A cold email sent today has zero value tomorrow. A useful email course sent today is still doing its job in 2027. A good newsletter issue is forwarded for years. A talk at a regional society meeting becomes a referral that becomes a pilot that becomes a case study that becomes a citation in another society’s keynote that becomes three more pilots.
The startups that build curricula early end up, twelve to eighteen months in, with an inbound pipeline of clinicians who already know them, already trust them, and already want to talk about pilots. The startups that don’t build curricula are still cold-emailing.
What this means for fundraising
There is an underappreciated point here: investors who know clinical healthcare love this. A clinical AI startup with a growing newsletter, an active email course, a list of clinicians who voluntarily ask for the next issue, and a calendar of speaking engagements at the right conferences is doing something investors can see, measure, and underwrite. It is the closest thing to product-market fit signal that exists in a market where the actual sales cycle is twelve months long.
Investors who don’t know clinical healthcare will sometimes dismiss this as “marketing.” That is information. It tells you which investors to take meetings with.
How GluonLabs thinks about this
This is exactly why we added Go-to-Market & Audience Growth to what GluonLabs does. We watched too many clinical AI startups build genuinely good products and then fail to find buyers, not because the product was wrong, but because the channel was wrong. They were trying to sell a peer-bought product through a vendor-bought funnel.
Our work on the GTM side is the same shape as our work on the engineering side: discipline, structure, and a refusal to take the lazy path that doesn’t compound. We help clinician founders launch the email course, write the first issues of the newsletter, build the case-study cadence, and identify the small conferences where their buyer actually pays attention. Then we help them turn the resulting pipeline into pilots, and the pilots into case studies, and the case studies into the next round of fundraising narrative.
A medical AI startup is not like other startups. The work is different, the team is different, and the way you find your first customers is different. Founders who internalize all three structural differences early build companies that compound. The ones who don’t spend their second year fixing the decisions they made in their first.
If you’re at the beginning of that journey — or the middle of it, and something doesn’t feel right — we’d like to talk.
GluonLabs is an engineering and product studio, not a regulatory consulting firm. We design and build clinical AI systems with regulatory frameworks (FDA, IEC 62304, ISO 14971, HIPAA) shaping our architectural and process decisions. Final regulatory strategy and submissions should be reviewed and signed off by a qualified regulatory professional.