Four Places We're Pointing Capital

Earlier this year I said we'd start with the problems our operators are actually trying to solve and let those become our investment themes.

This is the result.

Over the last 6 months, I sat down with service-line leaders, pharmacists, clinicians, IT, finance, supply chain, and asked the same question every time. What are your big messy problems? I wrote down every answer. There are now over 400 of them, each one classified by clinical domain, business lever, solution type, and which part of the health system owns the pain.

A list of problems isn't a thesis, though. So I ran every one of them through three filters, and four areas survived.

What the inventory told us

The demand is operational. Cost and efficiency is the lever our leaders cite most, then quality and safety, then strategic capability. Access and growth, the thing everybody talks about, is fourth. That matches what Galen Growth sees globally, where the operations-and-workflow layer took 46% of 2025 funding in the AI operating layer of care delivery, more than any other category by a wide margin.

Half of the problems identified aren't for a startup at all. Roughly 200 of the problems have what we call an "internal capability" tag, meaning the fix is mostly process improvement. Strip those out, ask which problems a start-up could actually solve and you're left with 121.

The three filters

Is it defensible?

The 2026 market is a recovery that's really a concentration. Digital health raised $4.0B in the first quarter, the strongest Q1 since the pandemic peak, but twelve financings of $100M or more took 59% of it. Bessemer has AI at 55% of all health-tech funding, and Rock Health retired its "AI deal" category this year because the label stopped meaning anything.

We read that two ways. We can't compete for mega-deals and shouldn't try. Our lane is $250K to $2M, pre-seed through Series B, exactly the stage the squeeze leaves behind. And "it uses AI" is no longer a thesis. When software is close to free to build, the software isn't the moat. So every pure-software company that comes through now has to show us two of four moats:

1. Proprietary data
2. Amazing product insight
3. Distribution or network effects
4. A regulatory barrier

Does our own strategic plan say anything?

We checked the inventory against Ballad Health's FY27–FY28 plan, and I learned it was not super helpful. By the time something is in an annual plan it usually already has a solution attached, which makes it a poor place for venture. What the plan is good for is telling us where the organization has already committed dollars.

Will Epic ship it?

This is the filter that matters most. I mapped every problem against Epic's shipped and future roadmap (at least what I think it is today).

Across the whole inventory, 47% of problems are structurally clear of Epic. But the pattern by function is where it gets useful. Clinical care and quality is 84% contested. Revenue cycle, 79%. Patient growth, 72%. Invest there and that startup is competing with a free incumbent. Workforce, staffing and supply chain are about 70% clean. Regulated devices sit outside Epic entirely. And the work of proving compliance to regulators and auditors is something Epic has not built and shows no sign of building.

Run the 121 through that filter and 39 come out both high-applicability to venture and outside Epic's scope. Those 39 are the purest ground we have.

The resulting four focus areas

I'll do a full post on each over the next few weeks. For now, the shape of them.

Focus 1 - Regulatory and compliance

No industry on earth is more regulated. U.S. hospitals spend $39B a year on the administrative work of federal, state, and other organizational compliance. For an average 161-bed community hospital that's about $7.6M a year, $1,200 for every admission, and 59 full-time people, more than a quarter of them clinical staff pulled away from patients.

We don't mean the compliance itself. We mean the labor of proving it: audit automation, evidence assembly, monitoring, and increasingly the job of governing the AI that other vendors are selling into the system. Our own leaders want help in their respective areas of responsibility.

Good problems here: governing and drift-monitoring third-party AI once it's in production. 340B and ship-to-account compliance. Credentialing, enrollment and licensing that takes months per hire. Quality-measure evidence that runs alongside the EHR instead of buried inside it.

Focus 2 - Med-tech for our core service lines and rural care

Med-tech is entering 2026 at a three-year high in venture activity, and the FDA issued 324 AI/ML clearances in 2025. A clearance is a moat by construction; nobody fast-follows quickly.

What we mean is regulated devices, sensors and diagnostic AI that enable service lines to change where care is delivered, improve quality, reduce cost, or improve the patient experience. We're most focused on the service lines where Ballad is growing: cardiovascular, cancer, neuroscience, urology, orthopedics, and general surgery. Those are the lines where our leaders named the most specific gaps and where extending a scarce specialist into a county that has none changes what care is possible there.

Focus 3 - Pharmacy

Pharmacy is a vital part of any health system's portfolio of services.

We wrote about this in June (https://www.balladventures.io/thoughts/the-prescription-is-the-relationship) A patient can fill a prescription almost anywhere now, and we said the local health system pharmacy should win that choice, not because we say so but because of what it can do that the alternatives can't. This focus area is about what has to be true for that to happen. Capture and refill that's easier than the mail-order app. Specialty pharmacy that's faster, more personal, and closer than the out-of-state fulfillment center. Prior-auth that speeds a fill instead of stalling it. The tools that make Ballad's pharmacy the obvious, better choice.

Focus 4 - The entry point to cancer care

Our strategic plan names cancer as a growth priority, and it's where our demand signal runs deepest by a mile: 41 oncology problems in the inventory, 13 of them high venture-applicability.

Within cancer, we're most focused on the front end. How does a patient with a scary scan, or a primary care physician with that patient in front of them, come to think it's a no-brainer to send them to Ballad? Right now the answer is fragmented. Screening is underused, referrals leak out of market, intake is scattered across sites, and the time from consult to first visit is longer than it should be.

Good problems here: screening that finds the cancer early and brings the patient back for the next scan. Referral tools that make Ballad the easy choice for a referring physician. Intake and navigation that gets a newly diagnosed patient to a first visit fast, with one clear front door instead of multiple.

What we're not claiming

These four are a focus, not a closed list. Workforce capacity and staff safety both surfaced with real commitments behind them, and we're watching them. Behavioral health and the AI operating layer that orchestrates other vendors' tools are live in the inventory too.

A demand signal isn't a guarantee of a customer. What it guarantees is that somebody inside the system has already said the problem matters. In my experience that's the difference between a pilot that runs and one that quietly dies in month four.

What I'd want a founder to bring us

Three things.

1. The operator problem you solve, in the words an operator would use.
2. Which two of the four moats you hold.
3. Why what you do sits outside what Epic will ship.

We write checks from $250K to $2M. What we bring besides the check is a live health system to validate in, operators who've already named the problem, and a rural, community footprint the rest of the market is now treating as its stress test. More detailed posts on each area are coming. If you already see your company in one of them, don't wait for it.