Latest / HumanifAI HealthTech / What Healthcare AI Is Missing at the Point of Care w/ Oron Afek, Co-Founder & CEO, Vim
Transcript
- 0:00Brian: Clinicians spend roughly two hours on EHR data entry for every one hour of direct patient care.
- 0:08Brian: That's this year, 2026.
- 0:12Brian: AI can and does solve that for doctors, therapists, and nurses, but the problem is getting the tech to where it needs to be.
- 0:20Brian: In a survey from the Digital Medicine Society, 72% of healthcare leaders said workflow integration is the single biggest barrier to AI adoption ahead of training gaps and leadership waffling.
- 0:34Brian: Today's guest is fixing exactly that.
- 0:37Brian: His company, Vim, builds the rails that carry AI straight into the point-of-care tools where clinicians already are, so they can give those two hours back to patients and not paperwork.
- 0:56Brian: Welcome back to Humanify Health Tech, where we explore how AI is writing the rules of healthcare and putting people back at the center where they belong.
- 1:05Brian: I'm your host, Brian Chee, senior director of content at Net Health.
- 1:09Brian: Today, I'm joined by Oran Afek, co-founder and CEO of Vim, a digital infrastructure company that clears the path for value-based care by connecting physicians, payers, and developers in a frictionless, AI-powered ecosystem that delivers real impact at the point of care.
- 1:28Brian: Before Vim, Oran served in the special forces of the Israeli military and built four companies in telecommunications, gaming, real estate, and education.
- 1:39Brian: Interestingly, not healthcare.
- 1:42Brian: He says he went into healthcare to make an impact, and as such, built Vim to run less like a typical health tech vendor and more like Uber or Amazon.
- 1:53Brian: Oran, thanks for joining us today.
- 1:56Oron: Thanks, Brian, for having me.
- 1:57Oron: Excited to be here.
- 1:58Brian: Good to have you.
- 1:59Brian: We-- You've said, you've said that you've built Vim to run less like a typical health tech vendor and more like an infrastructure other people build on top of.
- 2:09Brian: What, what did that instinct come from?
- 2:11Brian: What was the-- what was it about healthcare specifically that made you think that was the missing piece?
- 2:19Oron: Yeah, it's a good question.
- 2:20Oron: I think the, the initial incarnation of Vim started in back in twenty fifteen, and the original premise was, you know, we would start with a patient-facing intervention.
- 2:33Oron: Uh, we actually called it BookMD be-before Vim was, um, even the formal name.
- 2:38Oron: And we allow patients to select care, um, in places that are more, um, I would call them high-value, which is lower cost, better outcomes or similar outcomes.
- 2:50Oron: And the premise was, if we can make it as easy as shopping on Amazon, people would change their behavior, you know, use a platform that can lead them to a three hundred dollar versus a three thousand dollar MRI scan.
- 3:03Oron: With the same, you know, five tesla, uh, MRI machine, with even maybe the same radiologist that, that reads the same image.
- 3:11Oron: You know, in the morning, they're doing this in a freestanding facility called Glendale MRI, in the evening, they'll do it at Cedars-Sinai for ten X the cost.
- 3:19Oron: Um, that original premise, we thought, you know, we start with modalities of radiology.
- 3:26Oron: We would then go and, uh, expand it to minor surgeries, surgeries, colonoscopies, and so on and so forth.
- 3:33Oron: But starting from a place where there's usually a abundance of supply, like imaging, um, very, um, low, um, variation, relatively low variation in outcomes and, uh, obviously very high variation of price.
- 3:51Oron: That's kinda where we started.
- 3:52Oron: So similar thesis to what Amazon started with books.
- 3:55Oron: Again, uh, inventory is not a constraint.
- 3:58Oron: You can just print a book on demand and just ship it out.
- 4:01Oron: Um, and, and there's a huge selection of books.
- 4:05Oron: So si-similar thesis to how we started.
- 4:07Oron: We knew that this thing would have means to an end, like once we are able to, you know, conquer imaging, we can go to other services.
- 4:15Oron: That was the thesis.
- 4:16Oron: It failed miserably And the reason that thesis failed or the experiment failed is that we've realized that patient, no matter if you empower them with online bookings and the ability to, uh, see prices up front, even to waive their copay, people would still gonna be following provider's advice on not just what to do, but where to go.
- 4:40Oron: And that's something private equity, uh, firms realized, uh, a long time ago when they started to fund health systems to start purchasing clinics because they knew that if those clinics can start referring to those health systems and bypass, you know, limitations such, such as stock and anti-kickback, uh, just through sheer ownership, they, they could really benefit from, um, you know, uh, m-more volume in more expensive settings and much larger negotiation power when insurance companies will pay them three or four X Medicare fee that would otherwise a free-standing facility or a small clinic wouldn't get paid for.
- 5:18Oron: So we realized that we can't really break referral patterns by facing the patient.
- 5:21Oron: A-And that's how we evolved into the physician side.
- 5:26Oron: That was the first pivot, the first major pivot.
- 5:28Oron: Um, the second major pivot happened when we realized that what we have done, uh, by in, in that case, was guiding referrals to go to the more appropriate specialist by surfacing an incentive and data about where the PCP should be considering referring the patient.
- 5:47Oron: And that's actually been pretty helpful in altering referral patterns.
- 5:51Oron: We realized that actually, we actually Stumbled upon, intentionally or unintentionally, on a key technology component everyone is gonna be needing, which is integrating with clinical workflows in a seamless way.
- 6:04Oron: And once we realized that, maybe just like Amazon realized that they're sitting on a lot of compute that maybe is required in December during Christmas, but in January, again, like they're not gonna con-consume all this compute, and y-voila, you have AWS, right?
- 6:19Oron: We decided that opening the platform for third-- third-party developers to then leverage everything we have discovered, that's gonna be the right way to go.
- 6:28Oron: And that's what Vim does today.
- 6:29Oron: Vim is a developer platform that allow you to integrate with any workflow, by the way, administrative or clinical, we support today most ambulatory, acute, post-acute EMRs.
- 6:39Oron: Um, we support other system of records such as CRMs, payer portals, uh, ERPs, uh, even a bespoke vibe-coded platform you built last l-last night, we can support it for real-time read and write, as well as to present a j- a, a bunch of UI interfaces on top that allows either apps or agents to run.
- 7:00Oron: So that's the story in a nutshell.
- 7:03Brian: So, so essentially, what Vim, what Vim does, reading this back to you, is it, it… What Vim does, it allows, it l- it allows vendors to live inside the provider's workflow and influence the care decision at the point of care, at the point where decisions are being made.
- 7:18Brian: Is, is that right?
- 7:20Oron: Yeah.
- 7:20Oron: I would say in the last six months, uh, the platform have been evolving to work, again, continues to support the point-of-care engagement when in the moment patient and providers are interacting, so that's majority of our use cases today.
- 7:33Oron: But we also now supporting other, uh, call them server to server or backend use cases such as, you know, someone calls in at 2:00 AM, they want to schedule an appointment, not a problem, we can support those type of use cases.
- 7:44Oron: Um, here's a list of 100 patients, reach out to them, call them, get them scheduled for, you know- Uh, an office visit, a wellness visit, um, prepare me for the visit itself, create, uh, you know, chart prep materials, um, you know, automate, automate the order, um, post, post, uh, post encounter.
- 8:06Oron: So we, we did expand the platform capabilities, but the point of care was the initial, you know, nucleus.
- 8:13Brian: So, so a patient, let's say, um, let's say, uh, a patient is at the doctor's at, at, at their PT, at their at, at a, at a, at, at a doctor's appointment.
- 8:22Brian: That doctor's using Ambient documentation.
- 8:25Brian: The notes that are being created are then almost instantly being, being referred through your system to, to inform on treatment and a decision, payment, all the things, all the things aligned to value-based care that need to happen in order to keep the treatment happening in real time and giving the provider the decision-making power at that, at that point.
- 8:52Brian: Then after that, to what you just said, it's about aligning the, the tactics of making those appointments and, and, and getting the patient to the next appointment, all through your infrastructure.
- 9:05Brian: Is that right?
- 9:06Oron: Yeah.
- 9:06Oron: These are great examples for building blocks that we have on our platform.
- 9:10Oron: There's obviously a lot more.
- 9:11Oron: Um, you know, th-think about, uh, care management teams, you know, peer site teams, MSL ACO site teams that are not the clinician themselves, but they do interact with patients and, or interact with other administrative or clinical workflows.
- 9:24Oron: And yeah, I mean, you can basically build whatever you want.
- 9:27Brian: So it's, it's, it's, it's a powerful connection tool the, that, that actually empowers this powerful technology that is already available for, for providers to use throughout that entire ecosystem.
- 9:41Brian: That ecosystem, by the way, is something that I wanna ask you about because you referenced it, you referenced it on a pod- podcast, and I wanna bring this up because I thought it was kind of funny, but also very true.
- 9:52Brian: Um, one of the things you talked about regarding sort of healthcare and, and the system in relation to the first wave of EHR digi- digitization was that health systems didn't design how care worked.
- 10:08Brian: They, when it came to EHRs, they just digiti- di- digitized the broken workflow that was already there, and you called that paving the cow paths, essentially taking a cow path, f- slapping some asphalt on it, and basically saying, "See?
- 10:24Brian: Isn't this better?" No, it's not better.
- 10:27Brian: So, so that, that doesn't solve the problem, right?
- 10:33Brian: But, but maybe AI does.
- 10:36Brian: Is, is, is AI in relation to what, what Vim does a way to actually, um, turn those cow paths into an, an actual regulatory path that works in a, in a real-time basis?
- 10:51Oron: In order for Uh, generative AI to, to generally help not just administratively, but also clinically, it needs to understand the broader context of, A, you know, who's that patient?
- 11:05Oron: What's the benefit design?
- 11:06Oron: What's the plan information?
- 11:08Oron: What's the broader clinical information that is available about that patient?
- 11:12Oron: A lot of those datasets could be assembled, um, you know, at, at a given moment of time, but need to be updated.
- 11:20Oron: Some of those datasets can only be assembled once you, uh, uh, justify treatment, which is the patient is actually scheduled and is about to get to the clinic.
- 11:29Oron: Um, and then you can start pulling data from the broader, uh, health information exchange or, uh, i-industry, right?
- 11:36Oron: So essentially, you need to make sure that the data is available, uh, and is standardized and down to a markdown version that agents can interact with in real time, and that the right users can access the right data for the right patient at the right time.
- 11:49Oron: So there's a complex entity store that needs to govern who can access what and under which permissions and who signed a BA with whom, whether it's a third-party developer or a payer or risk-bearing entity or Medicare itself, or, you know, the physician is seeing a cash-paying patient or someone that is complaining about a mental health condition should be hidden from the payer 'cause that condition is being paid for, um, with cash, 'cause the patient doesn't want this thing to leak to their employer.
- 12:16Oron: So there are a lot of permutations and limitations of how you need to, A, get data, sanitize data, making it available, and again, right data pushing to the right user at the right time.
- 12:29Oron: Once you have that foundation Then enabling things like agents to run data and start fulfilling things such as preparing a chart or suspecting for net new conditions, or finding a clinical evidence for a known condition from last year that you want to revalidate or creating a care plan, right?
- 12:51Oron: Those things become a lot easier.
- 12:53Oron: Now, the-- but still need to be done, right?
- 12:57Oron: The third portion of this is actually how do you, um, again, okay, fine, you read things, but then how do you, uh, write them back into the system of record?
- 13:08Oron: That's the regulatory piece, Brian, you're talking about, which is you need to make sure that the information, whether the information we found or the net new, the re-- uh, derivative that you've created, like a care plan, those need things to travel back into that EHR and, and be documented for the purpose of, you know, let's say, Medicare audit for other, other, other ONC-governed elements of how those EHRs were structured.
- 13:35Oron: And when you can do all three as, as we can do, I think you can genuinely and truly start benefiting from the value of AI, right?
- 13:43Oron: So think about, um, automation, building skills, um, you know, creating repeatable, repeatable intervention for, for, for the same patient or similar cohort of patients, uh, and do this in a way that is-- doesn't require you to write code.
- 14:00Oron: You can literally vibe code or use simple English to, to, you know, create those skills just like you would use, you know, Anthropic's Claude CoWork or one of those, you know, sim-similar solutions.
- 14:12Oron: Um, the problem we've been facing is that building everything I just said on a normal Fire infrastructure or HL7 infrastructure was not possible.
- 14:24Oron: Um, it wasn't possible because the APIs that were exposed and are exposed by different system of record vendors are not consistent, right?
- 14:35Oron: They're not consistent.
- 14:36Oron: They're very partial.
- 14:38Oron: Um, in places where, you know, they've been made available, they've been made available with a lot of hurdles, um, such as Epic, as an example.
- 14:48Oron: Um, and everything needs to be done through that vendor priorities and that such-- just such vendors, you know, kind of preferences and strategy and all that.
- 14:58Oron: So it, it, it did not create an ability to have a consistent way by which agents or apps can interact across the board and in a cohesive way.
- 15:06Oron: So you don't have enough context, you don't have enough way to execute, and you're just creating, and you're overburdening the providers.
- 15:12Oron: Again, you can automate things on their behalf, and you, you're still, you're still asking for more and more clicks.
- 15:17Oron: So what we've been able to do is to build a technology that is able to interact bidirectionally with those systems without reliance on the underlying system of record.
- 15:27Oron: Uh, and do that, by the way, in a way that doesn't use, you know, sort, sort of computer use agents or computer vision, things that would be by definition probabilistic.
- 15:37Oron: Do that in a way that is deterministic.
- 15:39Oron: Do this in a way that, um, essentially we learn the workflow, we then create a layer of APIs on top, and then we allow agents and app to interact with any clinical workflow consistently in a deterministic way, right?
- 15:51Oron: So it doesn't matter if you, you know, the hundred different EHRs in an ACO.
- 15:55Oron: Once an app is being built, such ACO can operate across all the different clinics employed or contracted, um, with the exact same app, with the exact same, uh, way of exchanging data and talking with a physician.
- 16:08Oron: So this component was the big, you know, kind of breakthrough.
- 16:12Oron: Um, and obviously everything on top is, is, is great, but this is the underlying, you know, kind of magic that we've been able to sort through and, and build.
- 16:21Brian: So this is the-- do, do you think that this is the-- this is the-- this is what gets value-based care to work, that gets it to actually become more effective in our current system, in our current American system of healthcare?
- 16:36Oron: I, I want to hope so, yes.
- 16:37Oron: I want to hope that value-based care can be AI first, can be AI native.
- 16:42Oron: Value-based care can be a paradigm where the humans work for the AI, uh, and, and not vice versa.
- 16:49Oron: I know it's a bit con-controversial, but let me explain myself.
- 16:51Oron: I mean that if the agents themselves are aware of what needs to happen, what needs to be, quote-unquote, maximized, right?
- 16:59Oron: Um, data is in one place, actions can be synchronously written back and forth from system of records, and then you have the clinicians themselves that are the judges, the, uh, the, the people with agency that decide that, "Okay, that's what these agents have recommended.
- 17:18Oron: Do I accept or decline it?" Just like you do with your core before it takes an action.
- 17:22Oron: I think that's the future of value-based care.
- 17:25Oron: Otherwise, it continue-- it's gonna continue to be a whack-a-mole game.
- 17:31Oron: Uh, and for the most part, people are gonna continue to try to see if they can squeeze more dollars from, you know, uncovering more conditions or gaming the HEDIS system, you know, one condition at a time, without genuinely adding value to the system itself.
- 17:45Oron: So, um, I think if you have AI tailoring the right intervention for the right patient for the right-- at the right time, whether they're seeing a physician or not, by the way, but then you have the physicians acting as the gatekeepers for clinical decision-making, I think can win value-based care.
- 18:02Oron: And to do everything I just said, you need to have infrastructure.
- 18:05Brian: Right.
- 18:05Brian: But-- And, and so you're saying that essentially, I'm not gonna fix the broken system.
- 18:10Brian: I'm actually gonna use AI to adjust it and then build something on top of that that's better.
- 18:17Brian: Because AI is at that point, native AI is at that point where, where it could actually, it could, it could actually, um, adapt to whichever system, like an Epic system or other systems or other, other, other environments, um, to, to standardize that approach and thus make value-based care possible a-across a broader, a broader swath of, of healthcare.
- 18:43Brian: Is that right?
- 18:45Oron: Yeah.
- 18:45Oron: I mean, I think it's a really elegant way to define this.
- 18:47Oron: If, if we're thinking about those systems, um, it's basically a bunch of UI in a piece of software in a database, in the most simplified way possible.
- 18:57Oron: And the UI is sort of a-- The way I think about it is, is an API for the human brain, right?
- 19:03Oron: It's essentially like, you know, it's not a very effective or, or… Sorry, maybe it's, it's effective for us humans because, you know, we, we think in a different way-- in a certain way.
- 19:11Oron: But, um, it's not the most efficient way for agents to navigate a piece of, uh, you know, a database, let's say.
- 19:20Oron: So by making-- building a layer that is agent-friendly on top of the system of records that circumvent the UI itself- Again, allowing the physician to stand there in the middle, be the gatekeeper for clinical decision-making, but do allow the agents to seamlessly interact both with the system of record, but also with endless other system of records, the SNF, the payer, the specialist, and so-- the patient, and so on and so forth, and, and, and get those things orchestrated I think that-that's where the magic happens, right?
- 19:53Oron: Like, the agent doesn't need to rely on UI.
- 19:55Oron: That's why there, as you said, there is something that need is being built on top, allowing those agents to interact.
- 20:00Oron: Eventually, the agents become the interface.
- 20:03Oron: At the end of the game, the agent become database of record.
- 20:05Oron: I mean, those system of records would be eventually replaced by agents as well.
- 20:11Oron: I mean, the databases in by itself, right, currently regulated and just like CRM systems like Salesforce, you know, uh, kind of build this really unique ways of storing data.
- 20:23Oron: Those things are evolved as well into agentic databases that can be, you know, again, if, if the agent itself is the one interacts with your wearables, with your behavior, with your specialist, with your sniff, with your payer, that agent become a lot smarter than the EHR, and that agent would replace the EHR eventually.
- 20:41Oron: And we are building our infrastructure to, to this type of future.
- 20:44Oron: This future, again, not gonna happen tomorrow morning, but it's gradually gonna evolve toward that era, that system of record's gonna be replaced altogether by agentic workflows.
- 20:56Oron: And, you know, again, we need to be nimble enough and flexible enough to, to support things, you know, in a, in a gradual fashion, and that's what we do.
- 21:05Brian: That is so exciting.
- 21:06Brian: It is so exciting to be at the cusp of that and, and, and listening to-- listen to you talk about what that future could look like.
- 21:13Brian: Um, okay, so let's go, but let's go back in time a little bit because I wanna talk to you about your reason for tackling, which is, which is probably the hardest problem in the world or one of the hardest problems in the world, the healthcare system in America, right?
- 21:27Brian: So you grew up in Israel under a single-payer system.
- 21:30Brian: Um, there were what?
- 21:32Brian: Two, three tightly conne-connected HMOs, right?
- 21:37Brian: Um, you have said in the past that the idea of a doctor there making a decision about full longitudinal data on a patient is unheard of.
- 21:51Brian: In the US, that's kind of the default, right?
- 21:54Brian: And that the, the, the dichotomy between what you grew up experiencing in Israel and what you experience in the US made you think, "I can make an impact by f- by being part of the solution when it comes to the way the US handles, um, healthcare and the system that drives it." Is that right?
- 22:18Oron: Yeah, yeah, yeah.
- 22:19Oron: For me, it was overwhelming, still, still very overwhelming to think about a healthcare system that is, um, forgot who said it, but, um, not caring, not healthy, not a system, um, that is on a fee-for-service schedule, uh, meets the patients usually, usually when it's too late, that prevention is enforced through, you know- One time incentives or, or some-something that is, you know, I mean, HEDIS to some extent is, is prevention, but it's, uh, it's very limited in terms of the holistic view, where retention spend of a patient is as low as a year and a half, two years.
- 23:02Oron: It's basically equal to the time you keep a job with a certain employer, and then you switch a plan and, you know, Medicare churn is, is pretty high as well.
- 23:11Oron: So Medicare Advantage, I mean, and then when you think about the long-term view of a patient, either with a, from provider lens or payer lens, there's no long-term interest in patient care.
- 23:22Oron: And in Israel at least, you have, like as you said, the three HMOs, Brian, but you know- The average retentions- retention rate of a patient is around seventeen years.
- 23:32Oron: So it's much closer to a Kaiser, which obviously this country is able to get better results at lower cost, um, than it is to a fee-for-service system.
- 23:41Oron: So, so what you have, what you have, eventually you have the smartest phys-- most capable physicians in the world and the best technology here in the United States meeting, um, fee-for-service schedules and one-time encounters with patients.
- 23:56Oron: Uh, it, it, it just, it just ends up being very expensive and not very effective, and that's why we're seeing subpar results to other systems.
- 24:03Oron: So for me, the, the thought was, c-can we, can we leap, you know, kinda-- C-can we fix this to make it seem more like a single-payer system, even though it's still gonna be running on a fee-for-service rails?
- 24:14Oron: Which, by the way, fee for, uh, single-payer system like in Israel do have their own limitations, right?
- 24:19Oron: I mean, the government, government, uh, decides what the capacity should look like, and, you know, they're only gonna buy the, the ninth MRI machine after the eighth MRI machine is booked for twenty-four/seven solid.
- 24:30Oron: And by the way, these are the numbers that Israel does, uh, at least used to be allowing, uh, to serve ten million people, like one MRI machine per million.
- 24:38Oron: It, it's pretty, it's pretty nuts.
- 24:40Oron: So taking the free market dynamics that made, you know, this country so great and but, but putting the right infrastructure in place to, to genuinely allow for the right, uh, you know, care under the right incentives to happen.
- 24:58Oron: Now, I can't do everything, right?
- 25:00Oron: If, if, if I could, I would, um, try to convince Congress to, to stop employer sponsored healthcare 'cause I think it's a disaster, and double down on ICHRA vehicles that allow for, you know, more of a state-based employ-- uh, state-based cov-coverage, which is gonna be more long term, right?
- 25:17Oron: But I can't affect those things.
- 25:18Oron: So at least what I could affect is putting all the players in the same infrastructure on a common infrastructure and allow them to play and interact in a way that is transparent and easy and simple without being constrained by things that should be commoditized, like workflow integration.
- 25:34Oron: And you said it like this is a, a big hurdle for AI adoption, for value-based care adoption.
- 25:38Oron: So that, that was, that was the thought.
- 25:40Oron: Today, I think we can leapfrog the single-payer system benefits and go into an AI payer system.
- 25:48Oron: And, and, and, and AI, the beautiful thing about AI, it doesn't get tired, and it can be very specific, and it can tailor to ind-individual.
- 25:57Oron: Uh, and in that type of world, medicine becomes proactive, uh, not reactive.
- 26:02Oron: And, and that's not something you have in Israel either, uh, or Canada or Germany.
- 26:07Oron: Uh, a proactive healthcare system that is having a constant conversation with the patient, loops in physicians and interventions when needed, but it's, it's definitely a, a, it's a, it's a different, it's a different paradigm.
- 26:20Brian: So es- e- essentially you're saying it's keeping doctors from making decision, decisions in that void, right?
- 26:26Brian: The void of information, the void of retention.
- 26:29Brian: They don't know who the patient is.
- 26:31Brian: They don't have the information about the patient.
- 26:33Brian: They don't know what the patient is able to do as far as treatment or the extent of their treatment, and they're making those care decisions, what's the term, sick care decisions, on that patient- Mm-hmm without having all that information at their fingertips.
- 26:47Brian: Essentially, Vim is trying to eliminate the void there so the doctor can make the right decision through u- through the use of AI.
- 26:58Oron: Yes.
- 26:58Oron: And ideally, the doctors are part of large organizations, and we do lucky to work with quite a few of them that are physician aggregators that know how to adjudicate, quote unquote, um, the complexity of the payer with the reality of care delivery on the provider side, and we service their infrastructure in order to, you know, make care delivery more seamless.
- 27:26Oron: And I think that's a good way of kind of finding an interim solution to this adversary payer-provider relationship that drove this, you know, single largest industry in the world.
- 27:39Brian: Yeah.
- 27:39Brian: Yeah, yeah.
- 27:40Brian: Okay.
- 27:40Brian: So, so you talked a minute ago about fragmentation, I think.
- 27:43Brian: You know, we, we, we touched on, we touched on the, the sheer amount of differences in the system, the payers, Medicare, all the insurance companies that, that, that, uh, that a health system the, the doctors have to deal with, right?
- 27:58Brian: Um, that, that makes this an exquisitely complex problem.
- 28:03Brian: W- what does that cost a practice on a day-to-day basis?
- 28:10Brian: Not necessarily just in terms of, you know, um, abstract thought, but like as an actual waste of time or, or, or in, in helping patients heal, right?
- 28:21Brian: What, what is that picture that-- What is the picture of what that fragmentation costs?
- 28:29Oron: I mean, we can talk about today's terms or, you know, three-year outlook terms.
- 28:34Oron: I think, um, in a world where I think every individual physician is, is signing a contract with at least ten different payers, and, and each one of those payers have at least three different lines of business, and essentially you have thirty different sub-payers, right?
- 28:51Oron: Um, I mean, that's, that's pretty fragmented.
- 28:55Oron: Each one of those payers have a different portal, usually.
- 28:58Oron: Um, so ten different portals.
- 29:01Oron: Um, you know, different rules to submit prior authorization And, you know, d-different ways to reach you, need to care for those patients, how you guide them through referrals, which specialist can you send them to.
- 29:13Oron: A lot of those questions, as you said, uh, combine this with the administrative burden of documenting things in a certain way to get paid.
- 29:20Oron: Um, you're looking at at least, um, half of the time is, uh, um, essentially invested in non-delivering value for patient.
- 29:31Oron: Uh, but just, uh, you know, that's, that's half of the time of, of those providers.
- 29:35Oron: And by the way, the physician himself are not the only one participating.
- 29:38Oron: T- you have the largest ratio of support staff per cl- per physician in the United States than any other OECD country, which is, which is pretty phenomenal.
- 29:48Oron: Take that and also count the externalizations.
- 29:51Oron: I'll give you one example of an externalization.
- 29:54Oron: Those externalizations are even more expensive than the sheer administrative cost in a given practice of a primary care physician.
- 30:00Oron: But an overburdened primary care physician practice, wh-when they need to refer you to an MRI or CT scan, um, they would oftentimes elect not to even handle the prior auth, uh, submission process.
- 30:16Oron: And the reason is that they're overburdened, and you have large academic centers that surround them, and they know that those primary care physicians are overburdened.
- 30:27Oron: And they came there, and they said,"We would take it away from you.
- 30:30Oron: We, we would take care of that submission of prior auth.
- 30:33Oron: Uh, we have a full-blown staff who's gonna call your patient, get the appointment scheduled, get the prior auth submitted, get it approved.
- 30:40Oron: We'll take all this for you, and we'll send you back the biopsy." If I'm a doctor and it's like I'm already overburdened, that sounds amazing.
- 30:46Oron: Let's do it.
- 30:47Oron: Um, what's happening at this time is that a patient leaves the primary care office.
- 30:51Oron: They're getting, uh, a phone number to call that academic center into their call center.
- 30:57Oron: That academic center would first schedule the patient for the MRI that, by the way, costs f- between six to ten times more than a freestanding MRI in a different facility.
- 31:09Oron: And again, just one, one, one modality.
- 31:11Oron: We can talk about every other modality.
- 31:13Oron: Um, they would first schedule the patient.
- 31:16Oron: They would then get them to sign a waiver that if this wasn't covered by insurance company, in retrospective, they need to pay it out of pocket.
- 31:23Oron: You know, everyone just signed those waivers bluntly, uh, blindly.
- 31:26Oron: Um, the third thing, they would then go ahead and submit the prior auth.
- 31:29Oron: In many cases, the prior auth is only approved after the fact, after the service was already done or just before the service is getting done.
- 31:37Oron: So they have really good kind of perfect timing for how to do that.
- 31:40Oron: The actualization of this is that we end up paying for those MRIs, you know, whatever, six to ten X more than we would pay in a freestanding facility just because our PCPs are overburdened.
- 31:51Oron: And I can give you examples that go into pharmaceutical products and surgery, where the admit, uh, the, the admit rights of certain providers are, and so on and so forth.
- 32:03Oron: It's, uh, the externalities of overburdening our providers is, is enormous, and that's contributes tremendously to the cost of the system.
- 32:13Brian: Wow.
- 32:13Brian: Okay.
- 32:14Brian: So let's talk about Vim a little bit and what Vim is doing currently.
- 32:16Brian: I think Vim has been around for what?
- 32:18Brian: About 11 years, I think.
- 32:19Brian: Yeah.
- 32:19Brian: Um, you have close to 100,000 providers across 10,000 practices that are live today.
- 32:26Brian: Um, as I, as I understand it, a develop- a developer can put a tool inside a doctor's EHR without having to sell that through to every single clinic.
- 32:37Brian: It works, it works-- The Vim works to put that tool, that developed, that vendor's tool in the workflow of that doctor's office so that they can use it in real time.
- 32:51Brian: Is that right?
- 32:52Brian: So what does that look like?
- 32:55Brian: If that's true, what does that look like for the provider, for that doctor, for that therapist when they, when, when that occurs?
- 33:02Brian: What does that actually, what does that actually-- what actually happens there?
- 33:07Oron: Yeah, for sure.
- 33:08Oron: Um, so let me take a step back.
- 33:10Oron: So today, if you wanted to deploy an application in a given EHR, you need to go use case by use case, EHR by EHR.
- 33:19Oron: So, for example, you know, Veeam has over a hundred different apps built on our platform.
- 33:24Oron: Uh, no EHR would allow me to deploy them through a smart or smart on FHIR rails into that s-specific practice because nobody would allow you to build an app store in a single app, right?
- 33:36Oron: Um, we have a full-blown app store that can be deployed.
- 33:40Oron: Um, so it's multi-use case on any EHR, and that's a big difference, right?
- 33:45Oron: To-- Go and try to deploy a smart on FHIR app with Epic for risk adjustment, um, whatever.
- 33:54Oron: Those forty hours, quote-unquote, that the, the net-network that is gonna take place is gonna likely take three, four, five, six months to accomplish.
- 34:02Oron: That's one site of providers.
- 34:05Oron: So one provider site at a time, one use case at a time.
- 34:09Oron: So our approach is all use cases at zero time deployed on any EHR.
- 34:17Oron: And that, that opens the opportunity for developers exponentially right now.
- 34:22Oron: So build once, deploy everywhere across the Veeam network or bring your own providers.
- 34:25Oron: We have really easy for developers to invite mentoring providers to the platform.
- 34:29Oron: From a physician standpoint, initially, it was, you know, pretty compelling and simple, right?
- 34:34Oron: I mean, they, you know, open up their EMR.
- 34:37Oron: We are context-aware, so we see which patient is in context, what the physician is trying to do, and sending different API calls or cues to our developers to say, "Okay, it looks like your referral management solution or your ambient scratch solution or, you know, your prescription management solution is a right fit for this moment for this patient.
- 34:57Oron: Go ahead and surface it here as a notification." What happened is that over time, you know, we, we had more and more developers joining the platform, and now we are seeing physicians starting to be overburdened with our own platform or too many notifications, too many different apps, too many different options.
- 35:12Oron: So we are, uh, in the, uh, process of introducing, uh, an AI assistant that essentially is an app that can communicate with all the different apps Uh, through a single interface, just like a co-work, you know, as, as an, as an analogy.
- 35:29Oron: And is able to, again, communicate with those apps on the provider's behalf and do things on the provider's behalf.
- 35:34Oron: Uh, and so the apps still remain, but the assistant becomes an interface that is, uh, allowing them to more easily navigate simple, simple actions.
- 35:43Oron: And, you know, you, you would see that happening in other-- I mean, it's already happening in, in a, as I said, you know, co-work or ChatGPT.
- 35:49Oron: They do allow you to communicate with your calendar, with your email, so that's not very different.
- 35:53Oron: I think you're gonna start seeing this really soon with Apple, and the new series is gonna allow you to communicate with different apps without going to the actual app.
- 36:00Oron: Uh, I think you're gonna start seeing more of that happening on consumer world.
- 36:04Oron: Um, so we were taking a very similar approach.
- 36:07Brian: So, like, so when I first started looking into this, I thought, "This sounds like a marketplace almost," right?
- 36:14Brian: And then I started reading a little bit more, and what I'm coming to, what I'm coming to terms with is it's more like a massive communications network where your-- where Vim is actually sort of the-- translating all of the various inputs and all the various, quote-unquote, "languages" that the different systems are, are, are using to standardize that language and to, to, um, to, to, to sort of create more efficiency as a result of that.
- 36:46Brian: Sort of, sort of that… Is that c- I know it seems weird to think of it as communication, but listening to you talk right now, you're talking about an AI assistant that essentially is connecting all of the dots in terms of this kind of, sort of, sort of app communication network.
- 37:03Brian: It's fascinating to me, really fascinating, that that is how we can use AI to, um, simplify healthcare, simplify the process of healthcare, which is what it really is doing, right?
- 37:16Oron: Understand.
- 37:17Oron: Uh, I think it's a-- it's, it's the right way to think about it.
- 37:20Oron: And, and when you have common infrastructure, common, common language, it can really allow for seamless communication.
- 37:24Oron: And those communication could be agents talking with other agents, with apps, with payer data, with the underlying EHR data.
- 37:32Oron: Um, a-and that's what AI's so good at, which is taking those unstructured instructions and act upon them in, in a way that is, um, that they can deliver value.
- 37:42Oron: But, you know, we're just-- what we do is, you know, we're putting the right, quote-unquote, "Evals in place" to make sure that when the AI interacts with, let's say, clinical data or clinical workflow, they do that in a way that is deterministic and, uh, and, and not probabilistic.
- 37:54Oron: So in a way that is safe.
- 37:56Oron: Um, we don't, we don't wanna have a… You know, if you book the wrong flight, fine.
- 38:00Oron: But if you, if you book the wrong patient for the wrong appointment and, you know, uh, it could be a disaster.
- 38:07Brian: Yes, indeed.
- 38:08Brian: Uh, you've used a framework in the past, I think you-- I think from an advisor, to describe, describe the near-term opportunity here for healthcare in, in regard to, in, in regard to AI.
- 38:20Brian: Uh, you call it offense instead of defense, right?
- 38:24Brian: Well, essentially what you said was… I've got it here.
- 38:26Brian: Hold on.
- 38:27Brian: Let me, let me, let me read this.
- 38:28Brian: This is important, right?
- 38:29Brian: "Making providers more productive.
- 38:31Brian: You're going on offense to make providers more productive.
- 38:35Brian: Order automation, workflow automation, summarization, ambient scribing, all of that is included.
- 38:41Brian: The defense side of that is catching provider mistakes before they happen, which is what Vim, which is what Vim does," right?
- 38:49Brian: That's all the AI that you, that you have in Vi-Vim working alongside a clinician at the point of care.
- 38:57Brian: Walk me through how that works.
- 38:59Brian: What i- Yeah.
- 38:59Brian: Is there a consumer-facing risk here in terms of that?
- 39:02Brian: Or, or w-where, you know, uh, how does that, how does that-- how in a real-time setting at point of care does that actually take place?
- 39:15Oron: Um, when thinking about defense and offense, I think about AI adoption and how use cases are being adopted, um, in healthcare, but also likely in other industry.
- 39:25Oron: The defense use cases are, uh, kind of the low-hanging fruit.
- 39:28Oron: It's like, you know, you're about to do A, uh, you know, try to consider B or C because A is really risky, and here's why.
- 39:36Oron: Um, so it's, it's more of a, you know, listening and proactively surfacing an option.
- 39:43Oron: Uh, offense is, is kind of, you know, thinking about more advanced use cases where the, uh, the, the doctor is allowing the agents to go and operate on the provider's behalf.
- 39:54Oron: You know, oftentimes it's not like, "Hey, go take care of Brian.
- 39:56Oron: Go talk with Brian.
- 39:57Oron: I'm too busy for Brian right now." Or just, you know, "Go talk with Oran.
- 40:00Oron: I'm too busy for Oran." It-- that's not that type of offense. That's, that's like five years a- five years away.
- 40:03Oron: Um, offense in a sense that like, "Here is what I think that we should do.
- 40:11Oron: Go and do it for me.
- 40:12Oron: Um, automate this order, get this prescription, you know, get this car being ordered, like it's covered by that patient's benefits." Um, that's, that's kind of how we think about, you know, defense and offense and, uh, and how, how the use case is being adopted gradually and, you know, you need to meet providers and health systems where they are, and you need to understand where in the adoption curve they're at and how much, how much they're willing to take on in terms of the behavioral change and, you know, that's the idea behind this.
- 40:40Brian: So you, you, you've said-- you've talked about impact, and I started, I started the podcast talking about how you went in-- you, you started four businesses, then you went into healthcare and sort of said, "Well, you know, that's interesting." Like, most people don't go into healthcare after being successful starting four businesses, right?
- 40:59Brian: I can understand why you do that now because, you know, you're a visionary looking for a solution, or at least you seem to me to be this incredible visionary looking for a solution and have-- and making an impact.
- 41:11Brian: Um, what was that impact that you were looking for, um, that, that became them?
- 41:18Oron: The, the, I think for me personally, and you know, my co-founder likely have, uh, hi-his own view, but for me personally, it was fi-fixing the injustice that I think is the, uh, you know, I think Look, I mean, as an Israeli moving to the United States, very grateful for, uh, a lot of the, you know, the tech industry, the, uh, the military industry, a lot of the things that have dev- developed in Israel that pretty much attributed to the United States being the, you know, the mentor, the shepherd, the f- the financier of a lot of those industries, right?
- 41:50Oron: So coming here thinking, "Hey, you know, there's maybe something I can bring to the table," because what really motivates me is this, I think, just pure injustice of how the system is structured and how people end up paying for this.
- 42:02Oron: Um, you know, when, when someone pays their deductible, and then they are signing up for, I know, a million dollar worth of service in a certain provider and saying, "Yeah, I already paid my deductible, so, you know, I'm not gonna pay the rest of this." You actually are paying for that.
- 42:22Oron: You're paying it through the next year's premium that is seven or eight percent more expensive.
- 42:28Oron: We employers pay for it, like we all pay for, uh, this expensive system, and I think that obviously that, that's unfair for, um, you know-- Again, patient affordability or lack of affordability is, is, is in, is a place that is, um, I think it's playing-- in my, in my head, it's playing into fairness, it's playing into, you know, human rights, it's playing into those type of things.
- 42:51Oron: So for me, coming in and trying to make a dent and try, try to, try to get impact that, that addresses those, those issues is, is really motivating.
- 43:00Brian: So 11 years in, if I were to ask you, in five years, what do you want-- what would you like to see?
- 43:08Brian: Where do you wanna be?
- 43:09Brian: What would your answer be?
- 43:10Oron: I think in five years, um, the first AI native health plan that thinks about both, uh, paying for patient care and rendering patient care, uh, is already using Virmen to, uh, to treat at least a million patients.
- 43:32Oron: Uh, and it's so efficient, and it leapfrogs all the other players out there.
- 43:40Oron: Um, you know, pe- people that build a lot of or, or, or combine a lot of bespoke solution that don't speak with each other that, you know, from an organization standpoint, they just have a hundred thousand people, they don't know each other, and they can't coordinate a single action, uh, properly.
- 43:56Oron: And, and now you have a team of hundred people that is super tight, that is leveraging agents and data to treat a million patients.
- 44:04Oron: I think that's gonna happen in five years, and, and I, I really wanna be enabling that.
- 44:09Brian: Oren, thank you so much for showing us all of this, showing us your vision, showing us where the-- where we're going in the future, and what that hopeful, what that hopeful, optimistic view is.
- 44:20Brian: Um, you've made me excited to see how, how that unfolds.
- 44:24Brian: I think that could make real change to, to a lot of people.
- 44:28Brian: Um, mostly the reason for that is, you know, it's putting tools that are already developed, that are good tools, that are transformative tools where they need to be so that providers, doctors can, can actually use them and make not only, not only their lives easier, but their patients' lives more healthy and a little bit more preventive, um, in terms of the care they give.
- 44:51Brian: Thank you very much for, for spending this time.
- 44:55Brian: Um, I could talk to you for another hour and learn more and more and more.
- 44:59Brian: I'm, I'm, I'm, I'm so grateful for, for this time.
- 45:03Brian: I know you're busy though, so the problems you're solving do not wait for anyone.
- 45:08Brian: We'll let you go and jump into our recap segment next.
- 45:12Brian: Thank you very much again, though, for joining us.
- 45:14Oron: Thank you, Brian.
- 45:15Oron: Thanks for having me again.
- 45:16Oron: It's lovely.
- 45:21Brian: All right, recap time.
- 45:23Brian: Here are the top three things I took away from my conversation with Oron today.
- 45:27Brian: First, distribution and depth are the same failure at two different stages.
- 45:33Brian: A good tool that never reaches a doctor doesn't help anyone, and a tool that's technically integrated into an EHR isn't the same as one that's inside the workflow.
- 45:42Brian: Vim's bet is it's solving both, not just one, is what makes the difference.
- 45:48Brian: Second, the industry's own hype cycle assumes healthcare behaves like a consumer app, and Oron says that's usually wrong.
- 45:56Brian: Most health tech gets funded and built expecting people to download it and form a daily habit.
- 46:03Brian: Healthcare is a low-frequency behavior.
- 46:06Brian: That mismatch is part of why so many well-funded tools never get used.
- 46:12Brian: Finally, Oron grew up i- under a single-payer system where a doctor deciding without full patient history was unheard of.
- 46:21Brian: In the US, that's the default.
- 46:23Brian: His bet is that value-based care only works once the technology and the distribution to reach a doctor exists together, not the technology alone.
- 46:33Brian: And that should do it for today's episode of Humanify Health Tech.
- 46:36Brian: As always, thanks for tuning in.
- 46:39Brian: If you're joining us on Spotify or YouTube, we'd love to hear your thoughts in the comments, and don't forget to leave us a rating while you're there.
- 46:47Brian: Once again, I'm Brian Chee from Net Health.
- 46:49Brian: I'll see you next time.