The Information Exchange: The O-PIMP Episode

We're back after a few weeks off, and we came in hot:* OpenEvidence landed its first B2B deal at Mount Sinai! So we talk about what happens when a PLG darling tries to grow enterprise muscles, the competitive landscape as they do so, and whether the real moat is product or sales. * Then we pivot to MEDVi and the New York Times piece that has everyone talking: can you really vibe code a billion-dollar telehealth business for $20K? (And should you?) * We wrap with some wonkery — the ONC is the ONC again, the O-PIMP is born, and I walk us through why the Henry Schein v. Vyne case might redraw the line on what counts as a health information network in America.And yeah, you’ll have to listen to understand why this exists (and why it goes so hard):Relevant Articles* OpenEvidence Announcements* Mount Sinai to integrate OpenEvidence AI enterprise-wide* OpenEvidence and Tandem Partner to Streamline Evidence-Based Prescribing and Prior Authorizations* OpenEvidence launches Coding Intelligence* Abridge and clinical decision support* OpenEvidence’s Gambits: Some analysis of where OpenEvidence might go after PLG from last summer, including the enterprise motion* From Alert Fatigue to Approval Fatigue: An oldie showing how it was always logical ambient scribes would infuse CDS, another attempt beyond the pop-up* When Horizontal Meets Healthcare: A piece about OpenAI and how their healthcare business model differs and threatens different players* One Copilot to Rule Them All: The copilot convergence, which OpenEvidence is now rapidly joining* Abundance and Agent: A discussion of how AI-powered software development’s marginal costs mean players sprinting to build it all* The PLG Trap: The OpenEvidence v Doximity cases shows how the sword of openness (PLG) cuts both ways, as we discussed on the show* How A.I. Helped One Man (and His Brother) Build a $1.8 Billion Company: The NY Times article on MEDVi* Healthcare at Internet Scale: An article from last summer about the OpenLoop lawsuit that mentions MEDVi* Rik Renard’s MEDVi post* Death to ASTP, Long Live ONC* The Battle for the Soul of HIE (or at Least the Definition)Chapters* OpenEvidence’s Enterprise Pivot (00:00) - OpenEvidence lands its first B2B sale with an enterprise-wide deployment at Mount Sinai, embedded directly into Epic. The crew unpacks the tension between product-led growth and enterprise sales in healthcare (BAAs, PHI access, institutional sign-off) and how this move finally brings OpenEvidence into UpToDate/Wolters Kluwer’s competitive set for real.* The Great Convergence: Scribes, CDS, and RCM Collide (04:28) - Abridge partnering with Availity and UpToDate, back-office co-pilots moving upstream, front-office co-pilots moving downstream. Brad flags clinical trials enrollment as the next obvious adjacency, and the group debates who wins the “wedge into the chart” race.* PLG in Healthcare and the BAA Problem (08:21) - Why product-led growth has historically been almost impossible in healthcare because of PHI and HIPAA. OpenEvidence may be the closest thing to a Figma for healthcare, but the harder, more valuable use cases require enterprise contracts — and that’s a different muscle entirely.* Vibe Coding the Roadmap: OpenEvidence’s Shipping Velocity (13:03) - OpenEvidence is announcing something major every month. The group attributes this to LLM-assisted development and frames it as a wake-up call: if you’re not adopting Claude Code or Codex-type tools, you’re behind. But speed cuts both ways — if anyone can build an LLM wrapper on the same corpus, is velocity a moat or a vulnerability?* MEDVi, OpenLoop, and the GLP-1 Gold Rush (18:40) - A deep dive into MEDVi, a telehealth front-end on OpenLoop’s white-label MSO infrastructure reportedly generating $1.8B in revenue. Brendan connects it to an older lawsuit alleging fraudulent oral tirzepatide marketing. The real question: when website creation, content generation, and national distribution all approach zero marginal cost, harm scales faster than regulation can respond.* AI Doctors and the Guardrail Question (25:51) - If the provider layer also becomes marginal cost — AI doctors on top of white-label infrastructure — the need for guardrails becomes existential. The group draws parallels to Cerebral, opioids, and the recurring pattern of technology outpacing oversight.* ONC Is Back: ASTP Reverts to Its Original Name (28:25) - The ASTP is reverting to the Office of the National Coordinator. The mission doesn’t change, and the pattern is familiar — Democrats expand, Republicans slim down. Pryce mourns the logo, reveals the internal Office of Policy is now the Office of Programs and Implementation (”the O Pimp”), and the group riffs on missed merch opportunities.* What Is an HIE/HIN? Vyne Dental v. Henry Schein One (34:50) - The episode’s deepest policy cut. Henry Schein withdrew from ONC certification, arguably to dodge information blocking. Vyne is trying to use the HIE/HIN actor definition instead — but that definition is famously ambiguous. Brendan breaks down the three exchange topologies and how the ONC preamble’s carve-outs could let nearly every network argue it’s not an HIE/HIN. A Maryland judge’s ruling in 2–3 weeks could reshape actor status for every clearinghouse, ADT network, and API platform in the country.TranscriptWe ran the transcript through an LLM to smooth it out. So it’s a rough approximation of the conversation (and in many cases significantly clearer than our rambling), but notably diverges from the word-by-word blows quite a bit.Brendan Keeler (00:00) Ladies and gentlemen, we are back. It’s been a bit, a few weeks. We got Pryce, we got Brad, we got Brendan. And we’re here to talk about — let’s start with some of the buzzy stuff. OpenEvidence making moves. Pryce, what are you hearing?Pryce (00:02) We back.Yeah, so diving right in, some of the news coming out of this week was that OpenEvidence has seemingly made its first B2B sale or engagement with Mount Sinai in New York. Apparently they’ll be embedding their application directly into the EHR, which is interesting.Makes sense from a workflow perspective, but interesting for a lot of the things that we’ll talk about shortly and how that allows these products to sort of work in and around each other or compete with each other. And that EHR at Sinai is Epic, if that’s not self-explanatory to anyone. So yeah, OpenEvidence to me is sort of the evolution of UpToDate and the idea that clinicians leverage it to see recent studies, verified research, get help in coming up with a care plan for the patients that they’re treating. And now they’re gonna have this AI tool, which Healthcare IT News is reporting as Mount Sinai’s first enterprise-wide AI deployment across clinical roles.Which was surprising to me considering I think they’re live with Microsoft’s dictation tool. But yeah, first B2B sale for OpenEvidence. It’s interesting, it’s cool. It also means a lot about where the market is headed, where they’re headed, what other companies who might not have had OpenEvidence on their radar will need to do in order to continually compete for physicians’ attention. And for value.Brendan Keeler (01:41) Yeah.Pryce (01:42) What do you all think about that?Brendan Keeler (01:43) What’s interesting is you said they’re competitive with UpToDate. They were doing a similar job to be done — they were providing clinical evidence, they were providing studies — but their go-to-market motion to date until this was totally different, right? They were ad-supported PLG. And so this is a new muscle that actually finally brings them into the competitive set.Pryce (02:06) PLG being product-led growth for those who aren’t like Brendan and Brad, always plugged into the…Brendan Keeler (02:14) You don’t love acronyms? We’re not going to just toss acronyms left and right? CDA, HL7…Pryce (02:18) I mean, well, those I do love because I understand them well. Yeah, so you’re saying OpenEvidence sort of went to market as like, hey, if you’re a physician, you should download this app. It’s free and useful to you, but I’ll make money off of ads. Whereas UpToDate was always — was that Wolters Kluwer? And it was a B2B sale? Is that right?Brendan Keeler (02:42) Yeah, you’re charging a couple hundred bucks a seat, enterprise contracts with a totally different motion that’s just hard, right? Like if you’re used to just saying, hey, pick this up, make it really frictionless for Dr. Smith — or actually you or me, like it’s actually not that hard, as proven by the Doximity lawsuit, for anyone to sign up for OpenEvidence and use it. You can just sign up.You do that and you’re building a very different muscle. What are ads? How do we do ads safely? How do we go to pharma and monetize via pharma? That’s what they’ve historically done. It’s always been a question of, okay, when they get popular enough, how will they lever over into enterprise markets to sell? Do they have the muscle to go to the CIO, CISO, chief medical officer and say, hey, you should rip out UpToDate and put us in? That’s been the question — do they have that sales motion?Brad Thorson (03:31) Yeah. It’s a really nice natural experiment of what is the right sales entry point for these AI-enabled enterprise products, because every time somebody gets a strong enough foothold with a large enough client, they are looking for — okay, yeah, I’m doing your ambient note taking, but now I need to look at, can I do ICD-10 and CPT code generation?It does feel like in the chart, OpenEvidence is competing against Wolters Kluwer / UpToDate, but they’re also competing against the scribes and they’re also competing against HCC coding applications.Do we have evidence of the RCM tools moving up into the chart? And then we have patient engagement and whether or not that’s going to continue to follow the patient into the visit. It’s really a question of who can sell and what partnerships look like.Brendan Keeler (04:28) Right. And this is the great convergence that’s been happening, right? You have Abridge partnering with Availity for prior auth and partnering with UpToDate for clinical decision support.So the convergence of all the jobs to be done for the provider — not just note taking. Abridge has expanded to be a full clinical co-pilot, but they’ve always been B2B. And then, like you’re saying, the back office co-pilots are moving upstream, the front office co-pilots are moving downstream, and now you have this heavy weight of the product-led growth market saying, actually, I’m gonna try and lever over and compete in the enterprise market. That’s gonna be challenging.Because they are well funded if nothing else, and they are aggressive if nothing else, and they launched multiple things. They launched prior authorization in partnership with this.Brad Thorson (05:12) I was just going to say, some tools may not even have an enterprise sales motion directly into systems of record or the care delivery organizations. Tandem is just going to go in with OpenEvidence.Brendan Keeler (05:27) Yeah, yeah. And I think the two things that come to mind for me — I don’t know if you guys have an opinion — but we think a lot about integration. I’m curious what their integration looks like with Epic and EHRs and if they need help with that. I mean, who could help them with that? And two, ads. They can’t possibly go sell to Mount Sinai and then be surfacing ads. Can they? But I don’t really see…Pryce (05:40) I don’t know who would help with that.Brendan Keeler (05:55) You would assume not, but…Pryce (05:57) It’d be sweet to be like in Epic and then if you click a button you can buy tickets to the next game that night, because I would. OpenEvidence is getting paid.Brendan Keeler (06:05) Yeah.Brad Thorson (06:05) Wait, they don’t even have to serve ads. I mean, who knows what the contract says, but all they need to know is that these podiatrists are commonly following this care path and ending up at a certain decision. That data enriches the free product that they have. And ultimately, pharma is looking for better attribution. Okay, I sent my reps into Sinai, and they’re telling me that I have to pay their expense bill of $20,000. How come my providers aren’t ending up on a care pathway that recommends my drug? I’m not saying that’s the motivation, but that’s a very easy product to build. And there’s going to be a lot of pressure from the people who are putting up the money for OpenEvidence ads to get better attribution.Pryce (06:51) Yeah.Okay, so I have two quick thoughts. One is just the product-led growth concept where OpenEvidence started by convincing — gosh, I feel like it was some absurd number — 30% of physicians in the country are leveraging it.Brad Thorson (07:11) Insane success. Would have never, ever, ever predicted it.Pryce (07:14) I think a lot about tools that want access to clinical data and how if they were patient-facing tools, then the patient would just need to authorize their access to clinical data. But guess what? No way in hell can you convince a million, 10 million, 50 million patients to download something.So the wedge of being like, well, my end user is the provider. And then once you have a couple of million providers using your software, then it’s like, go tell your organization to just put this in your EHR. And then to Brad’s point, you’ve got a business associate agreement set. Somehow maybe Epic and Oracle and MEDITECH maintain the rails of the clinical documentation or the RCM work because that’s what EHRs are so sticky for — really that enterprise business that they’re driving. But does the brain of the EHR and the insight of “If this care pathway, then this” and selling data to pharma or clinical trials start becoming a little bit more of a battleground?Brendan Keeler (08:21) What I would say is one tension here — when you are just doing clinical decision support, PLG, like PLG in healthcare is almost impossible because of PHI and HIPAA. You have this tension of needing a business associate agreement. So if I go to Dr. Smith, who’s part of Mount Sinai, and say use my tool — it’s just clinical decision support. Well, great, fine. I can pick it up and use it. And Doximity previously had success with that for just doctor social networking, right? And then some call/SMS stuff to prevent them from having to use their own phone line. Those are all non-PHI use cases. Well, now they’re tacking on PHI use cases. And that inherently has to be enterprise in nature because Dr. Smith, by and large, can’t sign a business associate agreement on behalf of Mount Sinai and give access to Mount Sinai’s patient data.That tension of the subjugation of providers to increasing layers of hierarchy in the consolidation of providers in America means that historically PLG products have struggled or just not been really possible. We all want the Figma of healthcare and now we have one.But really you’re just seeing them mature into enterprise because you have to do the harder, more valuable things. You kinda need business associate agreements and to sign agreements with institutions in America in terms of healthcare. And so that tension’s there. And I’m curious if they run into that as they try and offer these products and still maintain their roots as PLG, but also offer the really meaty stuff.Brad Thorson (09:55) I just want to circle back — I think Pryce just came up with two incredible business ideas. Pryce, I think you’re suggesting UpToDate should get into the IAS space and use patients as their sales channel to providers. Pull down your records and then UpToDate can tell you exactly what to drop into MyChart to ask for something.Brendan Keeler (10:23) Well, here’s the thing — OpenEvidence had patient-facing capabilities. They got rid of it. They got rid of it because it commoditizes themselves in some way when the patients can go and access that clinical information and it dilutes their ability to attribute ads and decision-making for pharma. And so I’m curious if they try and relaunch that in some capacity, but at the end of the day, we’re just sort of putting LLMs over the corpus of scientific information in America.But like if I’m a doctor and I can go just be a doctor or access it as a patient, maybe I do that, to avoid that. So it’s just a weird — that business model, that third business model of patient-facing clinical decision support, which I want for my potential thyroid thing. Like I want to go look it up and understand the same way my doctor does.Brad Thorson (10:52) Isn’t that what the economy is right now?Brendan Keeler (11:17) Can you have both? Can you have your cake and eat it too? They’re seeing if they can have their cake and eat it three in terms of patient-facing, PLG, and enterprise.Pryce (11:27) Eat it three.Pryce (11:29) Yeah.Brad Thorson (11:30) Also Pryce, we should expect an announcement shortly — if OpenEvidence is not working on it — that they find a clinical trials enrollment partner, right? Because that feels like the obvious next thing to slap on here. Hey, we understand the clinical pathways and the relationship to drugs. We can help get prior auths done so that patients get on these drugs and we can help with the discovery and development.Pryce (11:31) Yeah. Yeah.You know, we are taking a big leap here. Just to be clear, when I first read “Mount Sinai to integrate OpenEvidence AI enterprise-wide,” I’m like, so there’ll be a SMART on FHIR window in Epic where you can read about lymphoma, like recent lymphoma studies, right? But I’m expecting that they use this as a wedge to say, well, we don’t have this yet, but if you let your providers launch the SMART on FHIR app and they can read my content, but you also let me read the patient’s chart, then it’ll be easier for me to serve what’s relevant to you. And then if you let me read what the provider orders, I can make clinical decision support recommendations in real time. And before you know it…Like I said, we made a jump from right now — this has nothing to do with PHI and they’re putting it into an EHR — and we’re assuming that they’re going to start letting actual chart data influence their care provision and their clinical decision support and then eventually their models or their business motion. But yeah, I don’t expect that to happen next month. We’ll see.Brendan Keeler (13:03) That’s why they need someone who can help them do deep bi-directional integration work. Nadler, you know where to find us. I got one more thought here, which is they are shipping so much. They must be adopting LLMs to accelerate — not vibe code, but to accelerate development.Pryce (13:06) Yeah.Brendan Keeler (13:22) The marginal cost of development is dropping and you can see it with companies like this that you’re just like, how are they doing an announcement of this magnitude every month? And I think it’s a wake-up call for everyone in the industry that if you’re not adopting real Claude Code type products or Codex type products in your institution, you’re behind because you can really expand your roadmap as they’re very clearly doing. It doesn’t mean all these shots on goal are gonna work, but clearly they’re shipping a lot with their war chest of money. And I think it’s because behind the scenes they’re using LLM-assisted development practices.Pryce (13:58) And to me, that’s a double-edged sword. I’m far from an expert in training AI models and understanding differentiation there. But sometimes I hear about things like this and I’m a little bit like, well, if this is an LLM wrapper on the corpus of scientific research that maybe is otherwise accessible, how quickly could an OpenAI or a Claude or a Doximity replicate this? They can do all this so fast, and so can everyone else. What is the — this is like when Americans moved west and people were rushing into Oklahoma. The Sooners. Getting dysentery.Brendan Keeler (14:36) Yeah!Brad Thorson (14:26) Yeah, I mean… Manifest Destiny.Well, I think that’s why the sales nexus, the enterprise sales motion, is the important thing. Because once you have that space, once you’re in that workflow, it’s easy to bring partners along and just say, hey, we’re going to take care of this, because they have to fight for integration priority. That is the…Pryce (14:49) Yeah.Brad Thorson (15:02) The ability to quickly roll out a new feature means that sales is — this is great for me — sales is quickly becoming far more important than product and engineering. So you guys now work for me, which is great.Brendan Keeler (15:15) Shots fired. Shots fired. One thing there is — in the PLG market, “Competition is a click away” is the saying. And that’s true, but look at Google. On the corpus of information available on the internet, we can go build a search tool. And yet 80% of people use Google. And so that’s why they’ve always positioned themselves as the Google of clinical evidence.But I think they look at it and they’re like, s**t, Doximity is shipping exactly that, and others are shipping competitive products. So we need to lever over into enterprise so that competition is a contract away. Competition is a two-year, three-year contract away. That’s a moat. And that’s the moat that they jealously look at UpToDate every day and go, I wish we had that, to prevent the existential dread. So I think it’s astute to call that out by both of you.Brad Thorson (16:03) Here’s my galaxy brain take for this podcast. If anybody can build this — well, I am constantly frustrated with getting my GLP-1 from my excellent PCP and his very responsive staff, but it takes — there was nobody making sure that I’m getting that drug that is coming from their 340B pharmacy and making them a ton of money, except for me. And if I wasn’t invested in getting my healthcare from an unnamed institution here in New York City, it would be really easy for me to just go online and find somebody who has a consumer experience to do that.And it turns out we now know that the cost of developing that is $20,000 because the New York Times is giving MEDVi maybe too rosy of coverage for what they’ve built, but allegedly generating — was it $1.8 billion? Is that right? In revenue?Brendan Keeler (16:58) Yeah, yeah. But if you think about it, their operational costs are quite high. So MEDVi is one of many front ends on OpenLoop, a sort of white-label telehealth provider that allows for you to set up a provider practice of sorts with any branding you want.Their operational costs — a large part of whatever, if 1.8 is true — much of that is going to pay OpenLoop and pay for costs. But make no mistake, they’re making a bunch of money.Brad Thorson (17:31) Yeah, but they’re competing against brands. 30 Madison got bought. They’re brands who had to put together all of this infrastructure. We didn’t have the idea of a 50-state PC that was focused on telehealth. It took a decade of digital health pushing on that door to develop it. We didn’t have the idea of mail-delivery compound pharmacies. There’s all these things that had to go in place. And then the part that really blows my mind is the cost of development of the website. It’s a form, right? They’re gathering information, they’re shifting it over to OpenLoop who’s providing an asynchronous visit, and then a drug is showing up at the patient’s door. Everything is LLM generated. They’re not hiring actors.Brendan Keeler (18:15) Have you seen the form, by the way? It’s like, what’s your height? And then there’s six options or ten options that say 1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10. And then what’s your number of inches? And it lists them all out. It’s a UI atrocity. It’s like a war crime. So yeah, maybe LLM generated. Also not the ideal flow. But somehow, through a variety of tactics, converting people.But Pryce, you had something you wanted to say.Pryce (18:40) Well, I was going to say — yeah, we changed this from OpenEvidence at Sinai to “you can vibe code anything now, including a $1.8 billion revenue business.” Shout out CMS, ONC — you’re doing God’s work because apparently there’s a lot of organic competition that’s shoving its way into these software ecosystems with their elbows, saying I have the right to that data as well and I’m gonna use AI to build my product. I’m gonna use AI and what my product spits out. So it’s happening. That’s what we wanted. And in some cases it might be a little bit sketchier than in other cases as we shift to the MEDVi conversation.Brad, you were talking about PCs and I was just learning about PCs and what’s their counterpart?Brad Thorson (19:30) MSOs.Brendan Keeler (19:31) MSOs.Pryce (19:32) MSO. Yeah. So explain to me the organizational structure of OpenLoop. How are these physicians staffed, and is the founder of MEDVi the one who is staffing all of these physicians or just leveraging their services to create his pipeline of prescriptions?Brad Thorson (19:57) Yeah, I mean, think of OpenLoop — the front door of the MSO — as collecting a bunch of services that are needed to enable telehealth businesses, whether that’s licensing, an EHR or telemedical service, or the actual providers that are going to see patients. OpenLoop does not employ them directly. They contract with professional corporations.And then those providers are delivering care under their own discretion. And they will continue to become — it is equivalent to like, I open a coffee shop, I use Square as my POS, and then after I’ve done enough sales, Square Capital can make loans to me because they know what my cash flow looks like. They’re just going to continue to hang tools on their toolbox, and a question arises — at what point does the collection of those tools start to look a little bit questionable?Brendan, I don’t know if you want to jump in and touch on something you’ve written about.Brendan Keeler (21:15) Yeah, I mean, this article came out and I’m like, man, this name sounds really familiar. We don’t do a ton in the GLP-1 space at HTD, with a few customers historically, but yeah — I’m ready for peptides.Brad Thorson (21:26) You guys are just skinny by nature, okay?Wait, wait — that’s the other part about the story that was crazy, is the speed at which they could — okay, they hit GLP-1s and then they rolled out hair loss and they just looked at any other business and were like, we can build a cheaper mousetrap, and just rolled. It was crazy.Brendan Keeler (21:51) I think that’s the article I wrote last summer, because this lawsuit — like most things that prompt me to think about healthcare — a lawsuit came out against OpenLoop that alleged that they were masterminding this sea of fraud, these fraudulent websites across the internet that were spun up to prescribe oral GLP-1s. This particular compound is not oral. It can’t be consumed orally. Tirzepatide-megalutide… I don’t know. We can look at the article.Brad Thorson (22:18) It’s blue time.Brendan Keeler (22:19) But I am not a doctor. Not a lawyer. Not a doctor. And so I went back and looked at that article and lo and behold, one of the “shell websites,” “shell companies” that was named in the lawsuit — as in, OpenLoop set up a fake website — was this company MEDVi. And so we now know that’s not true, that there are one or two guys behind this. They are a real company making some amount of real money.But that means that the allegations about fraudulence need to be directed probably at MEDVi and not at OpenLoop. They’re not named in the lawsuit, but there’s a lot of chatter on Twitter and some good threads. Rik Renard posted a great post breaking down that, okay, they’re making real money, they’re real — but there are tactics, things, the way they’re appealing to patients, that could be deceptive. And that’s fraud. Fraud is not the existence or realness of a company. It is: are you deceiving the consumer? They could be breaking — many people are pointing out that they may be breaking FTC, FDA, and other regulations and laws.We’re not gonna adjudicate that. We’re not lawyers, we’re not judges. But there is plenty of banter about that out on the market today.Brad, what were you gonna say?Brad Thorson (23:38) It seems like one of the major issues is that many brands were marketing oral tirzepatide. The fact that the product catalog was the same across all these brands seemed like the part of that lawsuit that the prosecution was most focused on, because tirzepatide cannot be delivered orally as far as we know.Brendan Keeler (24:00) It’s not trisibuclopetide. I think you’re right.Brad Thorson (24:03) Why do I know how to pronounce these drugs? This does not need to be space in my brain.Brendan Keeler (24:08) I just think one takeaway, and this was in the original article, is that the power of the internet is the ability to reach national and international audiences for trivial amounts of money. The marginal cost of building global or national businesses has dropped dramatically. And so if you and I wanted to set up a business doing fraudulent behaviors, historically it was like, okay, let’s go prescribe ADHD meds en masse — but we’re just doing it in our geography. We’re bound by geography. In the age of the internet, you can spin up and reach entire audiences at a national scale instantly. And so long as you can utilize that potential audience and get a meaningful percentage of it, the potential harms you can do are much larger, much quicker.We’ve seen it with ADHD meds and Cerebral back in the day, we see it rear its head here again, we’ve seen it with the opioid scandals. We see this all the time.Brad Thorson (25:18) I think the difference is — the internet compresses the cost to distribute information. But whoever wrote the paper about the transistor that led to LLMs, we now have technology that greatly reduces the cost to generate content. And that collision of super easy to distribute and super easy to create the content to distribute — that’s what makes the acceleration of this business possible but also fascinating.Brendan Keeler (25:51) And think about it — if we have AI doctors, imagine there’s some doctor behind the scenes of OpenLoop, but now we don’t need that infrastructure because we have AI doctors and they’re fully legalized. This highlights the need for guardrails as we go down that path, because all of a sudden, can you spin up a doctor and can you spin up a website? Great, you can do the full chain for marginal costs.I don’t think I know the answer — potentially you guys know the answer — but it just shows that the harm can be done very, very quickly, very, very broadly.Pryce (26:23) Well, in Rik’s post, it says that in their software, you can enter a birthday of February 31st, and then the app will tell you you’re quite likely to hit your goal weight of 200 pounds, even if you’re 7’11” and 350 pounds. So it’s like Shaq hoping to cut himself in half.Brad Thorson (26:31) Thank God. Have you guys seen how thin Charles Barkley is?Brendan Keeler (26:44) Yeah.Brad Thorson (26:46) He looks great.Brendan Keeler (26:47) Good for him.Pryce (26:47) Well, he’s got the Ro.You know what else this is making me think of? That you can build things so fast — the law and the repercussions of fraudulent activity or poorly designed software and clinical consequences now lag how fast you can build it and distribute it. And it’s making me think of how executive orders keep on happening in our country. And then a year later, we find out — we can’t erase the last year, but that wasn’t, you can’t actually do that.There’s a lot going on in the world right now that’s like, boom, the internet plus LLMs lets people create so fast that we’re having trouble keeping it safe. And sometimes it’s awesome. It’s productive. It should be a release valve on the healthcare system, which is way overburdened. It should be deflationary. I was listening to the CityBlock CEO, Toyin, speak about this.Brendan Keeler (27:27) How can we keep up?Pryce (27:49) On the HTN podcast — you want to see these things be deflationary in healthcare. And instead, oftentimes they’re used by folks who are like, well, how quickly can we prop up a lot of profit? And so we’re seeing a lot of inflation instead of a release of the burden of not having enough providers in the country. Instead it’s like, well, for the providers we do have, can we help them up-code all the way? So the government owes everybody more money. Anyway, this is just another example of — I don’t think this is helping the crisis that we’re having.Brendan Keeler (28:25) I had posted nine months ago, twelve months ago, where I was just feeling that exact thing. It is far easier if you’re an entrepreneur to make money off of making new money versus reduction of cost. It just is. And so the technological innovations we see go towards activating healthy consumers and patients to do new things, to spend more money a lot of times.Even the patient empowerment stuff we see via the CMS HealthTech ecosystem — all of a sudden you’re activating the healthy middle that never engages in their health. That’s really good, but the short-term effects of activating this population to use more healthcare is gonna increase costs. More people that were just not doing things — which probably long-term is bad for their health — well, if they’re utilizing all their benefits and spending out of pocket on all these different things, that’s more cost short term.And so I got kind of depressed a while ago because of that, because I was like, well, how are we ever going to fix this? And maybe Chris Klomp is the answer. He’s being thrown around all over the CMS as the person who’s going to make things deflationary. And I do have faith in him.Brad Thorson (29:46) Yeah, we gotta give the law heads what they want. The people come here for the wonkery. Let’s get — is there a new acronym?Brendan Keeler (29:46) Let’s segue in, given the Chris reference there.For wonkery. Wonkery. Pryce, what are you hearing?Pryce (30:01) Well, we’re back. We are so back. We are the ONC again. The HHS, Department of Health and Human Services, has many offices within it. One of those is the Center for Medicare and Medicaid Services, which uses a lot of the U.S. budget. Another one — much smaller and maybe I would say scrappier — is the Office of the National Coordinator, which was first named the Office of the National Coordinator for Health IT when Bush W had created it with an executive order. And then the last Biden administration changed it to the ASTP. So if you’re familiar with that acronym, that was the Assistant Secretary for Technology Policy. I’ve got my ASTP name tag right here. Amazing logo for the ASTP.Brendan Keeler (30:47) Vintage stuff. Hold on to that. That’s gonna be worth a lot.Pryce (30:56) I got my vintage Redox swag.Yeah, this is gonna go on eBay in like 20 years. Holographic ASTP name tag.Brendan Keeler (31:04) If they were smart — look, they don’t have much budget. They get 50 million, ever since Bush. If they’re smart, they would have sold swag at the ASTP annual meeting. That was a golden opportunity. They know this is coming. They know they’ve been working on it. They could have sold it. Everyone would have bought it. Problem solved. Deficit solved.Pryce (31:18) Do we think — is it legal for government offices to become like a…Brendan Keeler (31:23) What has legal ever been a problem for this administration? We’ll figure it out in court. The t-shirts. Should we sell the t-shirts?Pryce (31:31) I was thinking we should get Dr. Keane to do a collab with Supreme or something. Just super expensive clothes with ASTP — or now ONC — on them. That would be great.Brendan Keeler (31:44) It’s like the original LLM era a couple years ago where the Pope had the puffy jacket on. You know, we should do that sort of thing, but ASTP.Pryce (31:52) Yeah. Super swag.So with the change of this office from its name being the ASTP back to being the ONC, what else is happening? They divvied up some responsibilities.Brendan Keeler (32:08) It’s really not that much of a change. What Micky Tripathi did in summer 2024 is he pulled together a few other divisions so he could do more with AI — basically that was his goal. He had the chief AI officer or something as part of his title or under him, similar C-suite type people within the agency. And now they’re kicked over to another part. So it’s reorged out, and their mandate and mission doesn’t change.Even the verbiage that Micky Tripathi added in that change is maintained. So it doesn’t change much. And historically, I wrote a post about this, but almost every head of the ONC has done a reorg or multiple reorgs. Usually Democrats have added more sub-agencies and things like that, and then generally Republicans, especially Trump, tried to slim it back down.So it’s not crazy. And if the pendulum swings and Democrats get elected, guess what? Maybe we see similar additions to the ONC. I don’t know if we’ll do a name change again. But the ONC, let’s go through it. Office of the National Coordinator versus Assistant Secretary for Technology Policy. The former is better. Objectively better. ONC versus ASTP.Brad Thorson (33:31) ONC is easier to say for sure.Pryce (33:35) ASTP way better. ASTP’s logo is way better. No offense.Brendan Keeler (33:38) Wait — hot take — you like Assistant Secretary for Technology Policy better than Office of the National Coordinator?Pryce (33:46) I always liked saying — I don’t know, I think those words roll off the tongue if you can say them quickly. I also loved saying, oh yeah, hanging out with Steve Posnack is like hanging out with the assistant to the Assistant Secretary for Technology Policy. You get a little bit of The Office vibe. So I’ll miss those jokes.Something else that happened — I’ve got a friend that works at the ONC and I found out that he’s within the Office of Policy. That was historically called the O-Paul, I think, internally. But they just changed the name of that to the Office of Programs and Implementation, which I think should be called the O-PIMP now. So they’ve changed that name and I’m a huge fan of that change if we want to start abbreviating it.Brendan Keeler (34:34) I think we should. This administration has also been really focused in the CMS world on things like ACCESS and fun backronyms. Why not the ONC? Let’s dial it up here.Pryce (34:42) Backronyms, backronyms.Pryce (34:49) What do we call it? Oh gosh, we’ll come up with a good name.Brendan Keeler (34:52) So we got a few minutes left. Do we want to cover anything else before we close out?Brad Thorson (34:57) I mean, the people are asking, what is an HIN? And I think we have to tell them.Pryce (34:59) Brendan, you wrote about this. Maybe you want to tell us what you wrote about and then I’ll start commenting on my thoughts.Brad Thorson (35:08) I’ll just throw this in there, Brendan. I don’t really care about the Henry Schein case. I know you’re deep into it, but when you create grids, I’ll read every one of those posts. So if you could just lead with the grids, I’ll fight through the rest.Brendan Keeler (35:09) Sure.Brendan Keeler (35:18) Okay, the Excalidraw — that’s the tool I use, it makes it look hand-drawn. Yeah, Henry Schein versus Vyne is just a vehicle and a vessel for a very important conversation, which is: what is an HIE?So why are we having that conversation when Vyne, a point solution, a clearinghouse for dental, and Henry Schein, an EHR for dental — neither of them on paper are what we think of when it comes to HIEs. Well, it’s because the point solution, looking to use the Cures Act information blocking rules against Henry Schein the EHR, was unable to because Henry Schein backed out of the certification program to ostensibly avoid information blocking burden.Pryce (36:01) So just to be clear, there was an EHR who was certified by the government. And that’s important. Lots of EHRs do that because oftentimes you need to in order to get reimbursement from CMS and things like that. Lots of reasons to voluntarily certify your electronic health record. But if you do so, you immediately become susceptible — you are considered an actor as a developer of certified technology that could commit info blocking, which is really broadly defined as discouraging the exchange of electronic health information.So unlike most other industries where competitors fiercely hoard data and don’t share it at all, within the healthcare industry, the Cures Act says you have to share that data because it’s the patient’s data and other providers need it. And so we’re seeing this expanding ecosystem of like, you better share that data or you better make APIs.And you’re saying Henry Schein, being a dental EHR, had pulled away from being certified. It was after this lawsuit was filed, wasn’t it?Brendan Keeler (37:04) Yeah, and that’s one of the things — I’m not a lawyer, but what seems obvious is that the behaviors that were information blocking occurred last March 2025. They were a certified EHR. They were a developer of certified health IT — Henry Schein. And so you could claim that those behaviors were information blocking. Then they decertified, they withdrew in August or July. And the case was in September. So behaviors after that are nebulous. But it’s like, why didn’t you go harder in the paint on that one? Because that seems like slam dunk stuff.Brad Thorson (37:38) Yeah.Pryce (37:39) Right. Vyne, who’s suing Henry Schein, didn’t call it out.Brendan Keeler (37:46) Yeah, they did lightly in their opening brief and then in these transcripts they just talk about HIEs and HIEs. And we’ll talk about why they go there next, but instead — do the easy slam dunk, no nebulous definition route!Pryce (38:00) So within information blocking, there’s three different types of actors. These three types of organizations need to be very careful that they’re not committing info blocking. One being providers. Not really a way to skirt around that — if you’re a provider, you’re a provider. The second being a certified developer of electronic health technology, which would include a lot of major EHRs. And the third actor type, which is very nebulously defined — as Brendan loves to pull apart — is health information networks.Brendan Keeler (38:40) HIE slash HIN.Pryce (38:41) Okay, so the point being here, Vyne might have missed an opportunity to say, well, obviously they were an actor who could commit info blocking because they were a certified electronic health technology. And now they’re sort of saying, well, maybe they’re an HIE. This is an EHR that they’re saying was acting as a health information exchange. Let’s go into the details — what makes an HIE? Tell me about the preamble and all of that stuff, Brendan.Brendan Keeler (39:09) So essentially when you’re exchanging data, there’s only three patterns. One is decentralized. Everyone’s a node on the network and you send from point to point. So that could be a network that sets rules for exchange between their participants — could be a health information network. And many of those types of networks — DirectTrust, Datavant, Carequality — those are all point-to-point between nodes. Those could be.Pryce (39:36) Meaning all of the nodes trust each other, but you don’t get all of the healthcare information in the world and put it in one centralized server. Instead, the network is that you have a phone book, basically. And everyone in that phone book has a cert that proves, I’m in the phone book and you can trust me.Brendan Keeler (39:57) Yeah, in the preamble they mention — they carve out and the ONC says, actually, if you’re doing bilateral exchange between two parties, then you’re generally not going to be an HIE/HIN. And so most of those networks with that topology, even if routing through a central thing like Datavant or something, say we’re doing bilateral exchange and we’re good. We’re not an HIN because we’re not storing the data. We’re not facilitating a fan-out.And so the other end of the spectrum is what we think of when we think of state health information exchanges — stored in a big database. Everyone pushes data up and then when they need it, they pull data down. That model — that’s what Henry Schein is actually saying. Well wait a minute. We, as an API provider with an API program — people push data up to us, different apps that don’t know each other, to a single provider’s system, and they can pull down the data between them. One app pushes up, another can access it.And so Henry Schein is saying, no, no, no, that model — we’re not an HIE/HIN either. But if that’s right, then state HIEs of that model aren’t HIEs or HINs. And so the only remaining class of network that would actually fall into the definition would be ones where you send up a query and it goes to many participants, right? One to many, inherently not bilateral.But then at the end of the day, you just keep squinting at this. And I bet you every HIE, HIN, EHR — they look at this and their lawyers say, we’re not in this definition. And that can be fine. But right now it’s very ambiguous, because if Henry Schein is wrong, then everybody is an HIE or HIN if you offer an API.Pryce (41:52) If you offer an API and have multiple parties connected into you and you receive or share or reconcile the data, yeah, I see what you’re saying.Brendan Keeler (42:03) Yeah, where is the line? I don’t like this one, I hate it, because it’s so infuriating. And so in two to three weeks, we’re going to have a judge make the first call on what’s an HIE/HIN in this case. And that’s what’s important. That’s when we know if non-certified EHRs get looped in. I swear to God, every clearinghouse, every ADT network, every network in America — their lawyers are hyper-focused on this case, because it could draw them into actor status.And if they’re not paying attention, they better be.Pryce (42:34) Do we think this judge who’s going to make this decision has clerks or is he or she themselves calling the ONC and being like, what were you writing about here? How should I interpret it? Is that something they do? Because there’s so much ambiguity.Brendan Keeler (42:49) Yeah, the judge is showing a really strong grasp of these concepts. They’re thrown all over the board. They are federal judges.Pryce (42:58) I think they’re subscribed to Health API Guy. It’s crazy.Brendan Keeler (43:03) Yeah, I haven’t seen any of those domains come in. But he shows a strong grasp. He’s showing a very traditional view on computer fraud and abuse, the CFAA, which is what Henry Schein is accusing them of. And he’s really diving into the bilateral thing when interrogating Vyne. But at the same time, sometimes judges go really hard in the questioning and then they go and read the statute and the guidance and stuff that the agencies have released.I don’t know if they phone them up. I don’t think they do that. I think they generally rely on public artifacts and the law. And they’re gonna make a call. And he could make a call that really redefines this line. So we shall see.Pryce (43:50) Well, one more question — how would that redefine the line? Is it now legal precedent, and would that only be in Maryland, or would that be federal? Is it more like, I feel more comfortable making a decision like this because this judge made that decision in Maryland? Or is it, no, I can use this as a legal artifact from now on?Brendan Keeler (44:09) It’s both, right? It is limited to whatever the district court covers — Maryland and maybe some other areas. And then if it’s appealed, you see broader applicability. If it’s appealed to the Supreme Court, that’s when you get pretty uniform stuff. But yeah, absolutely there’s regionalization.But just one precedent allows for someone to point from a favorable jurisdiction. Someone in California can point and say, hey, look what happened in Maryland, and we have laws that kind of overlay on that. They can’t point to it as pure precedent, but then what you’ll see is — Vyne filed in Maryland, not in Utah, which they could have filed in because both parties have offices there. And that’s a very strategic choice to use the Real Time Medical / PointClickCare precedent.So we could see similar dynamics if it breaks a little bit towards Vyne or fully towards Vyne.But we are over time, so I think we gotta wrap. This was a fun one. See you all later.Brad Thorson (45:08) Yeah, we’re gonna go back to our normal jobs. See you guys.Pryce (45:12) Happy Good Friday! See ya! Get full access to Health API Guy at healthapiguy.substack.com/subscribe