WellAI

Intelligent diagnostics powered by 30+ million medical studies. Doctors and business owners, sign up for your zoom demo here:

[WellAI @RealWellAI]

@xsgames_ Hi! 🙋🏽‍♀️🤗 We are WellAI, a Digital Front Door solution for doctors and hospitals. The latest Large Language Models are at the heart of this solution.

[Sergei Polevikov @AIHealthUncut]

If you’re not at #NYTechWeek, where are you? 😉🗽

Come tomorrow, Thursday, June 4, for a live recording of Digital Health Vitals with @AlexKoshykov, Stephanie Davis, and Guy Milhalter. 🎙️

https://t.co/Td6O0ToITv

Hope to see you there! 😄

Go Knicks! 🏀🔥

[Sergei Polevikov @AIHealthUncut]

What's Actually Under the Hood of OpenEvidence's AI

My technical deep dive into @EvidenceOpen, one of the hottest healthtech companies right now:

https://t.co/JB97mluKjD

[Sergei Polevikov @AIHealthUncut]

The Most Humble Leader in Healthtech.

I once criticized @Infermedica in one of my articles. Instead of trying to silence me or attack me, the management team used the information in my article to fix the problems I was pointing out. The CEO, Piotr Orzechowski (@_orzech), circulated my article inside the company, and eventually the problem was addressed.

How fcking rare is that? It shouldn't be.

This and much more in my and @AlexKoshykov's interview with @_orzech, founder and CEO of @Infermedica:

https://t.co/QkUFgoWfn7

[AlexKoshykov @AlexKoshykov]

We sat down with @_orzech, CEO of @Infermedica — one of the pioneers in AI-powered clinical triage and symptom checking
We discussed healthcare, startup survival, Silicon Valley culture shocks, medical liability, VCs, future of clinical AI. @AIHealthUncut
https://t.co/SUnClcd8BC

[Sergei Polevikov @AIHealthUncut]

An update on the @EvidenceOpen / @doximity saga, and the latest developments in the Evidence-Based Medical AI space.

Also, stay tuned for Part 2 of this two-part series, where I dissect OpenEvidence’s AI.

https://t.co/NqhyTOgKcU

[Sergei Polevikov @AIHealthUncut]

LLM/RAG systems cannot produce "I don’t know" answers on their own.

In my two-part piece on the @EvidenceOpen / @doximity saga (https://t.co/Rl3sFtzATv), coming tomorrow, Thursday, I address in detail why LLMs, and in OpenEvidence’s case, a RAG layer on top of an LLM, cannot reliably produce "I don’t know" answers by default.

In fact, by default, they are optimized to do the opposite.

OpenEvidence claims otherwise. I see four possible scenarios for how they may be doing it, but none of them are simply "LLM/RAG architecture." They would require additional guardrails, classifiers, retrieval thresholds, abstention logic, prompt engineering, or another system layered on top.

But that defeats the purpose of using LLM/RAG to its full potential.

[AlexKoshykov @AlexKoshykov]

Health2Tech in May: The Digital Health Circuit https://t.co/sGMnKPghkE

[Sergei Polevikov @AIHealthUncut]

Bias in healthcare AI funding is real. But the irony is that the outcome has been the same: ROI has been negative across the board, regardless of subsector. Let’s be honest: healthcare AI is a brutal financial proposition unless you have the ability to “pump and dump” like most large VC firms do.

[Sergei Polevikov @AIHealthUncut]

GLP-1s are not a weight-loss trend.

They are a cultural earthquake.

A few takeaways from our latest Digital Health Inside Out episode on #Ozempic, #Wegovy, #Mounjaro, #Zepbound, compounded GLP-1s, and the obesity economy:

🧠 #Obesity is not a willpower problem

The “outsourcing willpower to pharma” framing is provocative, but it can also be misleading. These drugs work partly because they change hunger, fullness, cravings, and “food noise.” That is biology, not laziness.

💉 Patients are not looking for shortcuts

One of the most striking points from Dave Knapp was that many patients do not ask, “Do I have to take this forever?” They ask, “How do I make sure I never lose access?”

⚠️ Side effects are real, but so is desperation

Nausea, GI issues, muscle loss, and long-term unknowns matter. But many patients are willing to tolerate a lot because, for the first time, something actually works.

💰 Access is the real scandal

If the drug works, who gets it?
The patient with great insurance?
The patient with $1,000 a month?
The patient using a telehealth startup?
The patient buying compounded versions online?

🧪 Compounding is not a cute workaround

Sterility matters. Batch consistency matters. Dosing accuracy matters. cGMP matters. This is not the same thing as mixing protein powder in your kitchen.

🏦 But let’s be honest

When the official system prices people out, the unofficial system gets built. That does not make it safe. It does make it predictable.

🧨 The uncomfortable question

Are GLP-1s helping patients reclaim control over their health, or are we building the next forever-rent business model in American medicine?

Maybe the answer is both.

And that is exactly why this topic matters.

Huge thanks to Beverly Tchang, MD, @ManOnThePen, Sean Sullivan, @AlexKoshykov, and everyone who pushed this conversation beyond the lazy “miracle drug” versus “cheating” arguments.

GLP-1s are a real deal. But the system around them may still be rigged.

🎧 Watch or listen to the latest episode of Digital Health Inside Out on #GLP1 on YouTube, Spotify, Apple Podcasts, or wherever you get your podcasts: https://t.co/Z5hnJAvNIM

[AlexKoshykov @AlexKoshykov]

In our episode of Digital Health Inside Out, we decided to go deep into the world of GLP-1s.
And wow… this topic is WAY more complicated than “just take a shot and lose weight.”
Experts: Beverly Tchang, @ManOnThePen, Sean Sullivan, and @AIHealthUncut.
https://t.co/9D8dDa6dGy

[Sergei Polevikov @AIHealthUncut]

I asked Vilnius 🇱🇹 one question:

Can we stop the next Theranos before it actually happens?

My answer is: yes. But only if we stop pretending that “innovation” excuses bullshit.

1. Vigilance.

Vigilance from employees.
Vigilance from patients.
Vigilance from the medical community.

Because regulators, especially the @SECGov, have failed our industry.

Over the past few years, I’ve built a network of accountable and responsible people who are vigilant inside their organizations and inside their communities. I count on these people to be vigilant and to report the truth. These people are true heroes.

Companies love scaring employees with NDAs, compliance rules, and legal threats. Obviously, compliance matters. But people also need to realize something very basic:

Being silent about unethical behavior is being complicit.

2. Financial transparency.

Europe has many problems. But one thing I like is that, in most European countries, private companies are generally required to file financial statements. Switzerland is the big exception.

Now imagine if @theranos had been forced to file real financials.

Imagine everyone finding out in 2010, not in 2015, when a journalist, not the SEC, uncovered the 700M fraud, that the company’s ARR was closer to 100K, not 100M, as Elizabeth Holmes claimed.

The fourth-largest fraud in history might have been prevented.

Imagine Hippocratic AI being valued based on actual financials. Maybe 200M??? Not $3.5B because a couple of VC bros decided the spreadsheet needed more cocaine. (Speaking of @hippocraticai, the company appears to be imploding. Stay tuned. Some really ugly news may be coming...)

Transparency does not fix everything. But it makes fraud harder. It makes hype more expensive. And it gives employees, patients, journalists, clinicians, economists, regulators, and investors something to point to before the whole thing explodes.

Thank you, #Vilnius, for the brilliant and active discussion. 🙏

This is a global problem. Solving it requires global people who care enough to give their best.

Thank you @health2tech_.
Thank you BISEB.
Thank you European Humanities University.

[Sergei Polevikov @AIHealthUncut]

🚨 April was a crazy month for #healthtech.

Payer earnings. @WHOOP. @verilyhealth. @EvidenceOpen. @Talkiatry. eMed. @Click_Tx. CMS ACCESS. AI scribes and billing. @hippocraticai exiting international markets. ChatGPT for Clinicians entering the race.

So, naturally, our usual crew — @AlexKoshykov, Stephanie Davis, @BenSchwartz_MD, and I — had some calm, measured, totally unemotional discussions about what all of this means for the industry.

Just kidding. 😏

There were heated debates, thumbs up, thumbs down, accusations of delusion, arguments about whether "Series G" should even exist, and the eternal digital health question:

Is this innovation, or just another way to make the broken parts of healthcare bill more efficiently?

🎧 Check out the latest episode of Digital Health Vitals on whatever podcast channel you use: https://t.co/GLCdqMk8Um

And if you want to see the crew live, come see us on "tour":

📍@health2tech_ in Boston — May 26 (https://t.co/GdltOz0k0u)
📍@health2tech_ in New York City — June 3 (https://t.co/zuKWcyP5Ei)

Digital health may be confusing, but at least the debates (maybe???) are getting better. 🔥

[Sergei Polevikov @AIHealthUncut]

🇱🇹 #Vilnius! Here we go again.

On Wednesday, May 6, @health2tech_ continues its series of healthtech events in the greater Vilnius area, and I’m presenting again, this time on:

🚨 "How to Spot the Next Theranos in Healthcare AI"

I’d like to thank our brilliant local partner, BISEB, located on the business campus of European Humanities University in the heart of Vilnius.

🤖 What this is about:
The healthcare AI market, with a U.S. focus, and how to identify problems inside healthtech startups when press releases, investors, and much of the mainstream media are busy hyping AI up. I’ll draw on my years of experience as a healthcare AI fraud investigator. 🕵️‍♂️

🤖 What this is NOT about:
I will not be talking about AI models in healthcare. That’s a fascinating topic, and one I enjoy tremendously, but it’s for another time.

🤖 Who should attend:
Startup founders, investors, regulators, compliance professionals, economists, and anyone working in or around healthcare AI.

Please register here. And don’t forget to share with your friends. 👇

https://t.co/Vzo6Ti5Exv

[AlexKoshykov @AlexKoshykov]

Digital Health Vitals, Ep 6 is live 🎙️
This one had everything — earnings, fundraises, AI debates… and a few strong opinions 😄
Joined by @AIHealthUncut, @BenSchwartz_MD, and Stephanie Davis
https://t.co/aOpGIBDsZM
#DigitalHealth #AIinHealthcare #HealthTech

[Sergei Polevikov @AIHealthUncut]

@chrissyfarr is world-class. What struck me most about this interview was how unsentimental she is. She didn't romanticize her journalism years. She didn't pretend Lifers is what @CNBC was. She didn't deflect on the @a16z question. She didn't soften the labor-displacement critique. She didn't pretend @talkspace at 3.6x is exciting.

A founder writing the honest investor update about a failed EHR migration. The AI-native virtual care company that's profitable on day one because it didn't need to burn $200M to figure out unit economics. The scribe company that ignored the U.S. enterprise market and built a global business no one was watching. That's the version of healthcare Chrissy is rooting for. And it's the version a lot of people in this industry should be paying more attention to.

And if Chrissy is right about the future of AI-native care delivery, the next vintage of digital health winners isn't the next @wearehims or the next @HingeHealth. It's a generation of leaner, AI-native, vertical care-delivery companies that may never even need the public markets.

https://t.co/aO0mJ3lORO

[AlexKoshykov @AlexKoshykov]

Our interview with @chrissyfarr is live! We talked about:
– healthcare AI hype
– why most founders are terrible storytellers
– IPO reality vs. private market
– and why transparency in this industry is still… optional
https://t.co/xAIrZfrYCr
P.S. It's @AIHealthUncut's bday today!

[Sergei Polevikov @AIHealthUncut]

I'm happy to announce that I'm partnering with HLTH Inc., the organizer of two of the largest conferences in healthcare, HLTH and ViVE, on a critical content initiative called HLTH Voices, alongside renowned healthcare thought leaders such as @jareddashevsky, @nikillinit, and @GaryMonk. I'm humbled to be in the company of such celebrities.

[Sergei Polevikov @AIHealthUncut]

Former employees of @CarbonHealth are finally speaking out after the bankruptcy of what was once one of health tech’s most promising startups. There are some critical lessons here. And one especially important one: Carbon’s story was never just about Carbon. It reflects the deeper failure of a broken primary care reimbursement system.

https://t.co/5vL0yYBwfh

[Sergei Polevikov @AIHealthUncut]

We go inside the world’s largest AI scribe. “AI is underhyped” is one of many refreshingly honest statements from @AbridgeHQ CEO @ShivdevRao.

https://t.co/cxNrzVfq0x

[Calc Consulting @CalcCon]

Those who don't know, I was an NSF postdoc with @SchmidhuberAI PhD's advisor (Schulten) back in the 90s. 1 of 2 in the country. And my PhD groupmate recently won the Nobel prize for AlphaFold. So I have some qualifications here to say 𝐲𝐞𝐚𝐡 𝐭𝐡𝐢𝐬 𝐢𝐬 𝐩𝐫𝐞𝐭𝐭𝐲 𝐚𝐜𝐜𝐮𝐫𝐚𝐭𝐞.

The core learning principle behind JEPA is predicting one representation from another in latent space. And this was already explicitly formulated in the early 1990s PMAX work. PMAX does not merely hint at this idea; it sets up the same structure: two related inputs are encoded, and a predictor learns to map one latent representation to the other, while the encoder is trained to make this prediction possible without collapsing the representation.

That is exactly the defining mechanism of JEPA. When you strip away modern terminology and architectures, both are instances of the same objective: learn representations by maximizing cross-view predictability under constraints that preserve information.

What JEPA adds is not a new theoretical framework. It's just larger models, better architectures, and scaling. Of course, we could not do that in the 90s.

In that sense, Jürgen Schmidhuber made the real and original conceptual breakthrough: non-generative, latent-to-latent predictive learning

This is typical of @ylecun 's work; it's mostly derivative of others' ideas, scaled up and promoted. In contrast, @SchmidhuberAI really did pioneer a lot of these ideas. The JEPA work should have cited him.

Politics >> Integrity.