Rishi Khakhkhar

@ottosipe @doximity @photonhealth This is awesome man, congratulations!

Today @doximity and @photonhealth are launching Doximity Prescribe - free prescribing built into the app used by nearly every physician in America. I wrote about how our teams built this. photonhealth.com/blog/building-…

@nikillinit The rapid, accelerated aging in one of those groups isn’t enough?

need a different indicator for docs that completed med school vs docs that finished residency

In a recent @latimes op-ed, Dr. @HashemZikry, Medical Director at Counsel Health, explores the revolutionary potential of AI in delivering life-changing care at scale, and calls for serious, enforceable regulations to safely deploy this technology.

@anish_koka Random musing after watching millions of these clips: I feel like Jordan’s famous hangtime was less about his vertical jump (which was immense) but the amazing discipline/body control required to put up a good shot while he was descending instead of at his peak

Young people just don’t understand.

@dvasishtha True, though AI care delivery should strive to be much more. ChatGPT with a telehealth wrapper quickly, and rightly, commoditizes

Even if some of these "AI primary care physician" agents are incrementally better than chatGPT with a telehealth wrapper layered over, that service is actually desired by patients! It's seamlessly meeting them at their new entry point to the system.

People are already using ChatGPT like a primary care front door. This nature study shows that the 5% of LLM queries for health are symptom triage, “should I worry?”, “what do I do next?” happening outside the system. While there's a lot of pushback on the thesis of AI primary care we're Already seeing that: - Access has expanded massively (zero friction, always on) - But quality is uneven + unregulated - LLMs can have problematic behavior shifts (reassurance vs escalation) Fundamentally what this means is that some non trivial amount of primary is care being disintermediated at intake. And that AI native primary care that leverages the new front door has a chance to capture the top of tunnel while building for quality and consistency of care pathways.

@AdamRodmanMD @drjohnm Yup. @drjohnm, check out @PowerfulMedical’s Queen of Hearts mode! cc @RobertHermanMD

@drjohnm What's the algorithm? This is probably still a rules-based algo.

The way we are going…the perfectly pattern-recognition ECG read may be the last hurdle conquered by AI. It boggles my mind how bad these reads are. 2026 no different from 1990.

Agree w/ Mario. Frontier model improvement is great but writing evergreen, rigorous evals, optimizing latency, deciding what remains agentic vs deterministic, making sense of very messy health data — in short, major architectural work — takes a ton of energy and focus. I’m glad talented folks are working on this 😉

@DrDeepMD This is not even discussing the frontend interface, how the LLMs are securely managed, and any of the context layer for the agents to coordinate or many any tasks that engage with the data from the EHR/EMR or any other data that needs to be integrated from other systems.

This is much easier to build than we think We are half of the way there, building in simple easy hipaa compliant and BAA protected tools Clinicians should spend time thinking and relating, not being chart robots

Vibe shift: over the past few months, basically every computer in the community ER where I work has an @EvidenceOpen window up. Younger and older attendings alike. Huge usage everywhere.

Agree that we need malpractice/liability frameworks, but this has been done before (Waymo etc), I think we can solve that problem with the actuaries. Doctronic’s recent jailbreaking/prompt-injection incident raises the more pressing question of whether the states should be regulating this vs the feds. Should the bandwidth-constrained Dept of Commerce in Utah have been responsible for redteaming efforts? That seems unreasonable.

@tqlemd @Joshuabrowder I agree with this and you're already seeing versions of this template in Utah with doctronic (re med mal insurance)

My wife asked to schedule an appointment with her PCP and got a slot in March...2027.

The house of medicine is at a crossroads, I think the path forward is obvious

@dvasishtha We did this before! We villainized “Dr. Google” for 25 years and put our heads in the sand instead of building for the obvious access point of this century. AI care is coming, with or without us. We can help build it safely or accept the acceleration of our fading relevance.

If you as a physician do not believe AI should provide medical advice, here are a few things you should be advocating to the AMA and your local reps for to solve patient access issues: 1. Expand the federal residency cap - we've only added 1,000 residency spots since 1997 2. Fund more GME programs in community and rural hospitals 3. Expand pathways for international medical graduates and unmatched MDs 4. Support team-based care that extends physician capacity via NPs and PAs increasing scope of practice

@dvasishtha I think the same entities that oppose midlevel scope expansion will oppose AI doctors, for the same economic reasons :) That being said I do think AI doctors will be more rigorously scrutinized, evaluated, etc than current midlevel training programs and areas of scope expansion

If you believe that we're going to have AI doctors in the near to medium term future then shouldn't you also believe we can extend greater scope of medicine to physician assistants and nurse practitioners? Or that they'll have bespoke copilots to support expansion of scope.

@shyamalanadkat Go get em!

on the other side this time. different lens. same obsession with making AI useful.

@MarioATX_MD I see. Does the equation change now that the tech is more mature? We’re not there yet but feels very feasible that soon, a capable humanoid robot will be as good as a home health aide or hospital PCA for many use cases.

@khakrish Yeah, theoretically it makes a lot of sense. However, when you learn the real model for a lot of the leading caregiver services and even the facilities they partner with you’ll see that they don’t really care about the tech. It’s all about who will pay and how they get paid.

With MA crashing and Medicaid losing funding #1 is still not solved.

@RobertHermanMD @HenrytTimothy @JACCJournals @smithECGBlog @PendellM @belmontemarta1 Congratulations!

Queen of Hearts AI-based ECG analysis significantly improved STEMI detection, reduced false activations, and enhanced recognition of non-conventional presentations. Just presented by @HenrytTimothy and simultaneously published in @JACCJournals! #TCT2025

@dvasishtha Is it really Big Sean? Sign of a bull market

Excited to discuss how AI can support our neediest Medicaid beneficiaries from a bottle service table at the Big Sean concert this week! #HLTH2025

@CounselHealth is a full-stack, AI-native medical group that makes medical care abundant - unlimited guidance from real doctors, 24/7, scaled by Counsel’s medical AI, in a now familiar chat interface. This system embodies the qualities that everyone should expect from "AI as a New Site of Care", which go far beyond the simplistic, early models of digital health that simply replicated offline visits on a computer screen - e.g. care that is always on, with pan-specialty intelligence, personalized, and exponentially scalable. Thrilled to have led the Seed and be co-leading the Series A with our friends at @GVteam as Counsel launches to the general public. Counsel has served over 100,000 patients through B2B2C partnerships to date, and now everyone can have a doctor in their pocket. Congrats to @malagappan and the whole Counsel team on the launch!

Over 100 million people in the US are without a dedicated doctor. But it shouldn't have to be that way. The technology to scale medical advice exists today. I’m proud that @a16z is doubling down to co-lead @CounselHealth’s Series A alongside our friends at @GVteam. On top of that, Counsel's medical AI is now available to anyone. Our healthcare system is designed to ration access to our scarcest resource – doctor time. And in a world where clinician capacity is scarce, people are forced to turn to Google or ChatGPT for answers. Consumers ultimately need clinical resolution: the prescription, the lab test, the specialist appointment, the follow-up, the ongoing guidance and encouragement to change their lifestyle. Counsel is building healthcare abundance for the AI era. We led Counsel’s seed financing last year because of our conviction in this bold idea: that we can build a world in which everyone has their own doctor for life. They’ve seen immense progress under the leadership of @malagappan, who was previously a Stanford-trained physician, AI researcher, and healthtech executive. I’m delighted to expand our support as Counsel redefines how we access healthcare. @julesyoo @a16zBioHealth

Today we’re launching @CounselHealth to the general public. We believe Counsel is the most advanced combination of medical AI and human doctors in a single interface, and the easiest way to go from “I need advice” to “I’m getting care,” all from your phone. Learn more about Counsel ⏬