Half of athena's marketplace partners have joined since ChatGPT launched

Half of athenahealth's 500+ marketplace partners have joined since ChatGPT launched. I've tracked this ecosystem for 15 years and that number blew my socks off. But ChatGPT didn't do this alone. Three conditions had to be met first.

The platform had to exist.

athena's founder and CEO Jonathan Bush invited Harvard Business School professor Clayton Christensen to make the short trip from HBS to athena's Watertown headquarters around 2010. Christensen's disruption theory predicted startups would displace incumbents. Bush bet on a different model: open the APIs and let startups build on the platform instead of against it. athena launched its first marketplace in 2013 and rebuilt it with a new website and the MDP Labs accelerator program in 2017.

The regulatory walls had to come down.

Federal regulators who had paid doctors to adopt EMRs through the Meaningful Use program realized none of those systems talked to each other. The 21st Century Cures Act shifted focus to interoperability, forcing all certified EHRs to open specific APIs and stop blocking information. This was a place where athena was already investing, and it created the conditions for more healthcare technology companies to start building.

Demand had to spike.

Before the Cures Act even went into effect, COVID hit. Providers suddenly needed more technology fast. Telehealth, waiting room management, patient communications, just to name a few. Founders responded. But there was still a blocker. Developers were in short supply and commanded premium pricing.

Then ChatGPT removed the last barrier.

The platform was ready, the APIs were open, demand was there, but building health tech integrations still required deep technical expertise. LLM-powered coding tools changed who could build and the velocity of trained devs. The marketplace went from adding 72 new partners in 2024 to adding 128 in 2025, almost doubling in one year. The timing isn't a coincidence.

Most of these new companies won't make it.

Not because the technology is bad (although the idea of vibecoded apps handling PHI gives me nightmares), but because healthcare is operationally weird in ways that outsiders consistently underestimate. The workflows, the billing logic, the way a practice actually runs day to day, the regulatory layers, the fact that "the customer" is actually four different people with competing priorities: the provider, the office manager, the billing staff, the patient, PE firm, MSO etc. I've watched this pattern play out across dozens of marketplace partners. I've seen companies build technically impressive integrations and stall at 15 customers because they didn't realize that the nurse or MA, not the physician, documents key elements of the encounter.

The ones that survive will share one trait.

Genuine customer understanding. Not surface-level persona work. Deep, ongoing, "I spent a day in the practice and watched how they actually use this" understanding. And it never stops. The healthcare landscape shifts constantly with new regulations, new payment models, new patient expectations. Companies that stop learning drift into irrelevance even after early success.

The marketplace will keep growing. The technology barriers are gone and they're not coming back. What hasn't gotten easier is understanding how healthcare actually works on the ground. That's the real barrier now. And it's the one that no AI tool can lower for you.