I don’t do 20k steps a day. HIMSS made it worth it. Three days. 25+ meetings. One thing kept coming up that surprised me.
It wasn’t AI. Everyone’s talking about AI. It wasn’t interoperability. That word has been on every banner at every healthcare conference for over a decade.
It was this: “We built it. We can’t get it to the provider fast enough.”
Isaac Kohane, chair of biomedical informatics at Harvard, nailed the tension on the main stage: healthcare is going “too slow and too fast” with AI at the same time. Too slow – because outside of ambient scribes and revenue cycle, most health systems are still cautious about deploying AI where it touches clinical decisions.
Too fast – because Mehmet Oz, M.D., administrator for the Centers for Medicare & Medicaid Services (CMS) used his keynote to push for agentic AI for every Medicare beneficiary before this administration ends, and CMS has 700+ organizations signed onto its Health Technology Ecosystem pledge. The federal government is betting that AI agents can help seniors navigate care, manage chronic conditions, and choose the right health plan.
I’d say both things are true. And that’s actually a good sign.
At Vim, we focus on building the infrastructure that embeds health-tech solutions directly within the EHR workflow, so that the companies building point-of-care intelligence don’t have to rebuild for every system on the market. I spent most of my time at HIMSS meeting with those companies, across clinical decision support, revenue cycle, care coordination, and more.
The pattern across those conversations was stunning: innovation isn’t the bottleneck anymore. Delivery is. The gap isn’t intelligence. It’s access.
Almost every team described the same thing. They’ve built genuinely useful tools – things that catch a missed diagnosis code before a claim gets denied, that flag a high-risk patient at triage, that surface lab history without a provider leaving their chart. In many cases, the AI is already delivering real value. But getting that value into the ten seconds a clinician has during a visit, inside the EHR they’re already using, across the multiple systems a health network runs? That’s where it stalls.
This is the problem Vim is built to address. One integration, across many EHRs. The product is accessible within the workflow not after it, nor in a portal the clinician forgot they had a login for. When the infrastructure disappears, the intelligence finally reaches the moment of care. That’s not a technical nicety. It’s the difference between a product that demonstrates well and one that has the potential to influence real-world outcomes.
Three things I’m taking away:
- The “last mile” in health tech isn’t patient engagement. It’s a provider’s workflow. The companies gaining real traction are the ones meeting care teams exactly where they already work — not one tab over, not in a separate portal, not after the visit. KPMG’s Vince Vickers urged health systems not to get distracted by the shiny things when fundamental problems are right in front of them. He’s right. The fundamental problem for most health-tech innovators right now isn’t a better model. It’s reaching the care teams at the right moment.
- The best builders are workflow-obsessed. They don’t just know their product. They know the seven clicks that happen before and after their product shows up. That precision is what turns a good HIMSS demo into a product that care teams actually use.
- Collaboration is accelerating. CMS’s ecosystem pledge is pulling 700+ organizations toward shared standards. EHR vendors are signaling they’ll keep their platforms open for third-party AI. And the companies I met with see shared infrastructure as an accelerant, not a dependency. Build the intelligence, plug into the workflow, move faster together. That mindset is new, and it’s promising.
Quoting Isaac Kohane once again – Healthcare is moving too slow and too fast at the same time. But for the first time, all movement feels pointed in the right direction. The intelligence layer has been built. The governance is maturing. Now the work is getting it all into the workflow, where care actually happens.
