Shavan Patel

Epic Systems just announced they're partnering with Microsoft for ambient clinical documentation – and if you know anything about Epic's DNA, this is absolutely mind-blowing. We're talking about a company that has spent 46 years building literally everything in-house, from their own database engines to mobile apps, suddenly outsourcing one of healthcare's hottest AI capabilities. Here's what makes this so significant: Epic has always been the fortress. They grow their own food on their 1,670-acre campus, build their own everything, and famously told the world "if it's core to healthcare, we build it ourselves." Until now. The ambient scribing market exploded so fast that even Epic couldn't keep up with their traditional development cycles. Startups like Abridge and Ambience raised hundreds of millions while Epic customers were basically saying "give us this tech or we'll layer it on ourselves." Epic faced an existential choice: abandon their core philosophy or risk losing market share. Their solution? Partner with Microsoft's Dragon platform while quietly building internal alternatives – what I'm calling a "scaffolding strategy." Use external tech to buy time, then gradually replace it with homegrown capabilities. This isn't just about one partnership. It's a preview of how even the most successful tech companies will need to adapt as AI development becomes too complex and expensive for anyone to do alone. The implications for healthcare AI, vendor relationships, and market dynamics over the next decade are staggering. The full analysis dives deep into Epic's boardroom debates, the competitive shakeup hitting ambient scribing startups, and what this signals for the future of clinical AI through 2030. Link to the complete essay is in the comments. ___________________ Disclaimer: The views and opinions expressed in this analysis are solely those of the author and do not reflect the views of any employer, organization, or other parties. This content is provided for informational and analytical purposes only. Some of this content is speculative. It’s contents should be viewed as a mix of creative expression, facts and speculative storytelling. ___________________ Full essay linked in comments

334

44 Comments

How much does it cost to build a healthcare app? Short answer: $40k-$300k. Long answer: It depends on the complexity of its features, functionalities, and other variables. Here’s what actually drives cost for healthcare apps. 1. What problem are you solving? A symptom checker? A patient portal? A telehealth platform? The more regulated and high-stakes the product, the more complex (and expensive) it gets. Example: A meditation app might start at $60K–$90K. A HIPAA-compliant mental health platform with video + EHR sync? Easily $200K+. 2. Core features. Every feature = design + dev + testing + compliance. Want it HIPAA-ready? That’s extra architecture, encryption, and legal audit time. 3. Team structure matters. A senior team aligned from strategy → design → launch will build faster, test better, and reduce rework (read: you save money long term). 4. Launch ≠ The finish line. A scalable healthcare app needs maintenance, continuous compliance updates, feature iterations. Most MVPs miss this part. Want a ballpark tailored to your scope? We built a transparent cost calculator to help you estimate the project (link in the comments). ♻️ Share this if it was useful. 🔔 Follow Artem Borysenko for more updates!

49

61 Comments

Here’s your recap of last week’s health IT news 🗞️ 👇 🤖 𝗔𝗜 𝗖𝗹𝗶𝗻𝗶𝗰𝗶𝗮𝗻 𝗔𝘀𝘀𝗶𝘀𝘁𝗮𝗻𝘁 • Ambience Healthcare launched Patient Recap, a new clinical summaries solution. • Dartmouth Health implemented a custom AI model to triage and draft responses to patient portal messages, reducing clinician workload. • Arine raised a $30M Series C with participation from Kaiser Permanente Ventures and others. • Tampa General Hospital expanded deployment of Microsoft’s ambient listening technology for nurses via the Epic Rover mobile nursing app. • Abridge raised a $300M Series E. • Elation Health launched AI Actions to automate follow-ups, documentation, and care gap reminders directly within the clinician workflow. 📞 𝗔𝗜 𝗖𝗼𝗻𝘁𝗮𝗰𝘁 𝗖𝗲𝗻𝘁𝗲𝗿 • Infinitus Systems, Inc. + Salesforce deepened their partnership to expand agent adoption among life sciences and healthcare customers. • UChicago Medicine is deploying Agentforce by Salesforce to support operational workflows. • Commure launched Commure Agents, AI assistants designed to automate workflows such as scheduling appointments, providing patient updates, and managing referrals or prior authorizations. 🔁 𝗔𝗜 𝗥𝗖𝗠 • CodaMetrix announced a new “contextual coding automation” platform and an Emergency Department solution. • FinThrive launched Denials and Underpayments Analyzer, a tool to unify tracking and reporting on denials and underpayments. • Janus Health announced JanusIQ, unifying several of its existing solutions within a single platform. • VoiceCare AI raised a seed round of $4.54M. 🌡️ 𝗔𝗜 𝗥𝗲𝗺𝗼𝘁𝗲 𝗣𝗮𝘁𝗶𝗲𝗻𝘁 𝗠𝗼𝗻𝗶𝘁𝗼𝗿𝗶𝗻𝗴 & 𝗧𝗿𝗶𝗮𝗴𝗲 • Sutter Health partnered with Aidoc to access “earlier identification of critical findings, faster intervention, and better connected care.” • Arkansas Children's Hospital partnered with Caregility on NICU remote monitoring and digital health initiatives. • Valley Health System partnered with Inbound Health to launch an in-home care program. 📋 𝗣𝗮𝘁𝗶𝗲𝗻𝘁 𝗔𝗱𝗺𝗶𝗻𝗶𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 • Ushur launched Ushur Intelligence, a new suite of conversational and generative AI tools for patient engagement. • Altera Digital Health launched Sunrise CarePath, a patient engagement platform supporting digital intake, scheduling, and personalized communication. • Berry Fertility launched Smart Compose, an AI tool to assist patient support teams in drafting responses. 💼 𝗣𝗿𝗼𝘃𝗶𝗱𝗲𝗿 & 𝗣𝗿𝗮𝗰𝘁𝗶𝗰𝗲 𝗔𝗱𝗺𝗶𝗻𝗶𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 • Certify raised a $40M Series B.   ⚙️ 𝗖𝗹𝗶𝗻𝗶𝗰𝗮𝗹 𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻𝘀 • Emory Healthcare partnered with Aidin to coordinate post-discharge follow-up and manage referrals.

45

2 Comments

The agentic AI conversation is getting interesting, but I think we're missing a critical point. Everyone's talking about what AI agents can DO. Nobody's talking about WHERE agents should RUN. 🤔 Your autonomous trading agent managing millions in transactions - running on someone else's infrastructure? 🏥 Your clinical diagnostic agent handling patient data - processed by external providers? 🏭 Your manufacturing optimization agent with IP secrets - deployed on third-party platforms? The irony: The same enterprises that won't put customer lists on external platforms are building their most strategic AI agents there. 2025 prediction: Enterprise agent strategies will be defined not just by capability, but by sovereignty. The most valuable agents will be the ones you truly own and control. Are we overlooking the infrastructure question in our rush to deploy agents? #AgenticAI #EnterpriseSecurity #AIStrategy #DataSovereignty #EnterpriseAI

36

1 Comment

According to the 2025 HealthEdge Consumer Study, even though 64% of today’s members are either somewhat or very satisfied with their health plan’s ability to provide personalized experiences, members increasingly expect more individualized support, and health plans face challenges delivering that at scale. AI-powered tools offer a practical and scalable way to deliver better service without increasing operational overhead. While health plans have been pursuing AI-based innovations to automate and streamline back-office operations, such as claims management, provider network management, payment integrity, and more for years, AI adoption for member-facing engagements is still relatively new. Learn more about consumer AI sentiment and what health plans can do to provide increasingly personalized consumer experiences at scale: https://ow.ly/8j7H50WnJII

52

1 Comment

Healthcare has been held back for decades by closed practice management systems and data blocking. At NexHealth, our mission is to solve interoperability and unlock innovation for practices, developers, and patients everywhere. Synchronizer Wins is a new series that shows this mission in action. Our first story features Dentina.Ai, founded by a practicing dentist who refused to accept that nearly 40% of patient calls would keep going unanswered. With the NexHealth Synchronizer, Peter Gabbay: ✅ Launched an AI receptionist in 2 months ✅ Onboards practices in about an hour ✅ Enables 24/7 booking so patients never slip through the cracks This is what health data interoperability looks like when it works. Practices grow, innovators move faster, and patients get the care they deserve. 🔗 Read the first Synchronizer Wins story: https://lnkd.in/gZuxeEGW

74

4 Comments

CVS Health has the largest pharmacy footprint in the U.S. But what if its most powerful new location wasn't a building at all? What if it was the front-facing camera on 200 million smartphones? The current digital health journey is still episodic. A user opens the CVS Health app to refill a prescription, book a MinuteClinic appointment, or check ExtraCare deals. It's a powerful tool, but it's fundamentally reactive. It’s time to redefine the CVS app from a digital storefront into a proactive "Health Mirror." Imagine this: A user doesn't just open the app. They look into it. And in 35 seconds, using only their phone's camera, the app reflects their health, and magic happens. This is possible with binah.ai, which transforms any smartphone into a medical-grade vital signs monitoring tool. No extra hardware needed. Instantly, the entire CVS Health ecosystem becomes predictive, personalized, and perpetually present. How this redefines the customer relationship: Pre-MinuteClinic Triage: Before a telehealth call even begins, the patient's real-time vitals are already on the practitioner's screen: blood pressure, heart rate, respiratory rate, Hemoglobin, and even stress level. The appointment starts with data, not just symptoms. Proactive Chronic Care: A patient on hypertension medication gets a notification. Not just "Time to refill," but "Time to take 35 seconds to check your blood pressure and see how you're tracking this month." Adherence is no longer a guess; it's a data-driven feedback loop. Personalized Wellness Commerce: The "Health Mirror" detects a high stress level or low energy marker (via Heart Rate Variability). The app doesn't just sell vitamins; it recommends specific magnesium or B-complex supplements based on the user's real-time state, turning the retail aisle into a personal wellness consultation. Health Equity at Scale: This isn't a premium feature. It's an accessible one. You're democratizing baseline health monitoring for every single person with a smartphone, turning every home into a potential health checkpoint. By integrating binah.ai, CVS Health can transform its app from a tool people use when they're sick into a companion they engage with to stay well. The pharmacy is no longer just on the corner. It's in the palm of their hand, looking back at them. Let's discuss how to build the future of proactive, personalized healthcare. #CVSHealth #DigitalHealth #HealthTech #Innovation #Telehealth #PersonalizedMedicine #PatientEngagement #FutureOfHealthcare #AI

31

1 Comment

my takeaway from our healthcare data conference - the healthcare data people seem to be the most jaded about value-based care Some of the things I heard - Trying to do things that are critical for value-based care like attribution is basically impossible, so most arrangements are based on best guesses that none of the data people seem to feel very confident about - For most value-based care programs, the lion's share of money saved or made comes from documentation fiddling (e.g. risk adjustment) vs. actually changing the care model - The lag in data basically makes it impossible to actually forecast properly or intervene with patients at critical times you could actually help them Now I don't think most of the healthcare data peeps think fee-for-service is way better, I just think the optimism around value-based care is much more muted (at least in its current form)

281

34 Comments

No stellar candidate ever said, "I joined because their dental coverage included orthodontics." Yet I watch companies spend MILLIONS annually optimizing dental plans. Here's what candidates actually say: • "I can see exactly how I'll progress here - their career framework is completely transparent and rewards the people who actually deliver" • "A sabbatical for their best people? They're investing in me for the long haul, not just extracting value" • "Everyone I spoke with was incredibly open about challenges and growth - they clearly back the people who embrace both" These are Level 4 benefits. Cultural amplifiers. Competitive moats. The tragedy? Most companies spend 75% of their budget on Level 1 table stakes that every competitor offers. Stop Copying. Start Creating. Quick gut check for executives: What's ONE belief your company holds deeply that your benefits DON'T express? If you can't answer in 10 seconds, you're probably wasting resources on generic benefits. Example contradictions I see daily: → You say "we value innovation" but punish failure → You claim "we trust people" but require 3 approvals for a £50 course → You preach "we move fast" but everything requires committee approval and 6-week implementation cycles Your benefits should PROVE your beliefs, not contradict them. I explored this with David Duckworth and Carl Chapman on Ben's recent Friends with Benefits podcast, and I'm going deeper over the next 5 weeks. Starting tomorrow, I'm launching "The Belief-Based Benefits Series" - a complete framework for transforming benefits from cost centre to competitive advantage. What belief does YOUR company fail to express through benefits?

40

8 Comments

If your sales team seems to lack clinical knowledge, health systems will assume your implementation teams do too. If your product demo has the right features but nonsensical clinical content and workflows, they’ll assume you’ll mess up their specific configurations too. If you’re slow to respond in the sales process, they’ll assume your support team will be slow to respond to customer issues too. If your marketing exaggerates what your product can actually do, they’ll assume you will exaggerate your promises to them post-sale. If your salespeople care more about selling than helping them buy, they’ll assume you’ll upsell them on crap in the future. Why? Because the uncomfortable Truth is that culture is set at the top of the vendor’s organization. Whatever the leadership team is willing to tolerate in the company - even if it doesn’t happen all the time - becomes the “minimum allowable outcome”... which means it WILL end up occurring from time to time. Which is why as annoying as it might be to some people, at SeamlessMD, we’ve learned to be annoying about excellence, quality and being detail-oriented - across the entire customer experience from sale to implementation to on-going partnership. “It’s just a sales demo and the feature works… so why care if that one word has a typo?” is NOT an acceptable response. Why does that tiny little typo matter? Because if we have a culture that misses that one typo and doesn’t care… then maybe we will miss some other detail that will REALLY matter, that impacts the health system and their patients. But if instead we are a culture that is obsessed with excellence over the tiniest part of the customer experience, even if in the sales process - then that expectation of quality and attention to detail will permeate all the way to the most important thing we do: the patient experience. And health systems know that. Because they live and breathe in an environment where a missed detail costs lives. So Health Tech startups: if you want to earn health system partners, you need to drive excellence across the organization. You have to obsess over the details. Otherwise you may be short out of luck with health systems… and perhaps that’s fair.

45

18 Comments

Clinicians pivoting into Customer Success, here’s a project idea that could help land you interviews…💡 If I was pivoting from clinical care to CS today, I would do this one simple thing: analyze churn in my current role 👉 In business, churn is when customers stop using a product or service. 👉 In healthcare, an example of churn is when patients don’t come back after their first visit or drop out before finishing their plan of care. Here’s your project roadmap 1. Collect the data • Who didn’t return? • Who didn’t finish their POC? 2. Dig into the reasons • Cost • Didn’t see the value • Went to a competitor • Negative customer experience 3. Package the insights • Create a short report • Show how much revenue was lost due to churn • Recommend playbooks to prevent it Why this works: Leadership cares about retention because lost patients mean lost revenue and it’s more cost effective to retain a current client than get a net new client. If you can spot churn risk and propose solutions, you are already thinking like a Customer Success professional. Follow Lauren Braley PT, DPT, NCS. For more Non-Clinical career advice, I am a PT turned CSM for a health tech company and I have held several Non-Clinical roles, I help coach clinicians looking to pivot away from bedside 🌱

51

2 Comments

Why are QA-based benchmarks and single-turn evals bad? We tested GPT-5. In a QA format, we asked a model if hydroxyzine PRN can be prescribed to a 68-year-old patient with anxiety who uses oxycodone. The model correctly stated that it goes against AGS recommendations. We then presented a 68-year-old patient and instructed the model to conduct screenings and make a recommendation on whether a medication was appropriate. It was a multi-turn conversation, and the model managed to gather all the relevant information and then prescribed... hydroxyzine PRN. So, how come in a one-turn conversation the model recognizes the issue, but in a multi-turn it still fails? We don't know yet. We're still working on a mechinterp paper to answer this and similar questions. We see this behavior across many different use cases - it’s not only about medication or diagnosis. It generally seems that models perform way worse when used as a chat. The point remains the same. Using a QA-based benchmark to sell a conversational AI assistant is a dangerous manipulation. Evals 👏 for 👏 multi-turn 👏 assistant 👏 need 👏 to 👏 be 👏 multi-turn 👏 also 👏

34

3 Comments

Did you know that it takes $2 billion and 10 years to get a single drug to market? Even with ever increasing R&D spend? It’s time to rebuild biotech. #AI connected to the right data infrastructure is how we make discovery faster, smarter, and more successful.  Here’s what that means for the industry — and for the next chapter of Benchling: https://lnkd.in/gwrDzcRe    #RebuildBiotech

25

A while back, we released the Healthcare Agent Orchestrator - a solution template for building multi-agent systems in clinical workflows. Since then, we’ve been learning alongside customers how these systems can fit into the realities of healthcare IT. In our latest update to our Healthcare AI blog series, we dive into one of the toughest (and most exciting) challenges: connecting electronic health record (EHR) systems to a multi-agent, orchestrated AI system. We explore questions like: - How can multi-agent systems realistically connect to existing EHRs? - What integration patterns make sense for Healthcare Agent Orchestrator - direct SMART on FHIR access or mediated access via a harmonized data platform? - What are the architectural challenges and tradeoffs when bridging clinical data archives with patient record stores? Read the blog + check out the sample code (link in comments)! Kudos to Chris Burt, Josh Mandel, MD, Matthias Blondeel, Shrey J., Evan Miller, Manoj Kumar, Kumar Thirumalaiah, Frank Tuan, Yu (Aiden) Gu, Will Guyman, Alexander Mehmet Ersoy and many others!

90

7 Comments

Why do we hire neurosurgeons to do sales calls? How do you move fast in an industry where moving fast could harm people? How does every new customer onboarding fuel product innovation? I dove into these questions and more on the SaaSiest Podcast, discussing what it really takes to build trust and scale AI in healthcare. We covered why 80% of our commercial team has clinical backgrounds, how to build trust in high stakes industries and how every customer interaction teaches us something new about the real pain points clinicians face. Healthcare is not like other SaaS markets. You can’t break things and iterate when lives are at stake. But you can move fast and understand deeply. Thanks to the SaaSiest team for the thoughtful questions and great conversation! Link to the full episode in comments, worth a listen for anyone building in regulated industries or trying to scale beyond typical SaaS playbooks.

134

12 Comments

HMO to Fight Back AI UnitedHealth Group's earnings call yesterday had a few interesting points. One remark that caught my attention: "We are shifting to narrower networks and focusing on more disciplined managed products, particularly in Medicare Advantage." It seems the emphasis on HMO has already started. Looking at the enrollment trends across all UnitedHealthcare plans, the HMO line (shown in yellow) was gaining some popularity in 2024, but then took a sharp turn in 2025 and is now growing rapidly—more than ever before. Another interesting point from the call: "We had initially assumed Medicare Advantage medical cost trend of just over 5% when configuring our 2025 bids, in line with our normalized trend experience in 2024. We now expect full year 2025 trend to be approximately 7.5%." The situation is even more dramatic in Medicare Supplement plans: "Medicare Supplement, which typically represents overall care activity levels and cost trends in Medicare fee-for-service broadly, is up similarly in 2025 compared to historical levels. We expect that trend to be over 11% this year versus 8 to 9% in recent years—further confirming the broad-based nature of the care activity, and the coding and billing patterns trends we are seeing." Yes, aggressive coding activity and billing technologies were identified as one of the culprits for the MLR hit. Most AI technologies in healthcare landed in the provider sector first, with the promise of increased "productivity"—which on the flip side is somewhat synonymous with additional "revenue gain," which again involves "billing optimization." What I am getting is that, to fight back against these billing technologies, the current strategy seems to be closing off the gate and going to narrower networks—hence HMO. Interesting dynamics are playing out with AI technology potentially affecting this space. An arms race is emerging where providers leverage AI to maximize revenue capture, while payers respond by tightening network controls and reverting to more restrictive managed care models.

72

6 Comments