Gaurav Trivedi

@prathoshap Do you have / have plans to support Vedic accents too?

A 15-year-old dream has come true today. I started a PhD with the dream of creating a system that chants any Sanskrit shloka perfectly. And here I am opening sourcing 𝐕𝐚𝐠𝐝𝐡𝐞𝐧𝐮 - 𝐀 𝐯ṛ𝐭𝐭𝐚 (𝐦𝐞𝐭𝐞𝐫) 𝐚𝐰𝐚𝐫𝐞 ś𝐥𝐨𝐤𝐚-𝐭𝐨-𝐜𝐡𝐚𝐧𝐭 𝐭𝐞𝐱𝐭-𝐭𝐨-𝐬𝐩𝐞𝐞𝐜𝐡 (TTS) 𝐬𝐲𝐬𝐭𝐞𝐦 𝐟𝐨𝐫 𝐒𝐚𝐧𝐬𝐤𝐫𝐢𝐭. This is the world's first vrutta-aware, open-source TTS for Sanskrit Chanting.

@joshuapliu Abridge started with a patient app first and later pivoted to the clinician scribe, no?

Patient AI scribes will NOT take off as independent, sustainable businesses. I know, strange for me to say that, right? But hear me out… First, I know that sounds weird coming from a guy who literally started a Health Tech company that engages and empowers patients. But I’m also a guy who has spent 13 years understanding the gut wrenching, practical realities of what healthcare innovations do and don’t work as actual self-sustaining businesses. And the more I think about what it takes for a patient AI scribe company to thrive independently, the way that AI scribes for clinicians have spread like wildfire for companies - well I just don’t see it. With AI scribes for clinicians, you are starting with a massive, huge pain point. Some physicians have dozens and dozens of patient visits a day. Documentation is a huge recurring problem. When you have a daily, high frequency headache, the situation is ripe for a Painkiller you would pay for. But for the vast majority of patients, a doctor’s visit is a “once in a while” thing. Even for a patient with complex care needs, those visits are rarely daily. Yes it might feel like a huge headache for some patients in that moment, but the frequency won’t justifying searching for, signing up, and trusting a patient AI scribe even if it’s free - let alone paying for an AI scribe with money or your data (which let’s be honest, for those free AI scribes, you’re paying for it somehow…) And even for the subset of patients who want a scribe, most will just start by recording and throwing the transcription into chatGPT. Which means the market for patient AI scribes is just going to be too small to lead to a thriving business. The most likely way a patient AI scribes exists is if the scribe is a side feature of a more sustainable core product. Eg ChatGPT for Health just adds it to its core patient platform. So even if only 10% of users ever use it regularly, it doesn’t matter to them. But a patient AI scribe as a core product? As much as it helps patients, I just don’t see how it works independently to survive. What do you think?

I’ll die on this hill. AI models need to be trained at the individual clinician-level, not the population-level, in order to achieve widespread adoption. Physicians think about problems differently from one another, and do their work differently as well. A single AI model to be used “for all” will only be used by a minority of physicians and will be abandoned by the rest. Rather, AI systems need to be thought of as the individual clinician’s agent. And like a dog, I think that each clinician is going to have to spend substantial time working with and training the model to make it work in the way they need it to. Until I see this - a lot of this AI software is mostly hype to me.

The current use of AI in healthcare by patients and clinicians is not aligned with the evidence. In the new edition of Ground Truths

We are living through the disruption of healthcare in real time. Every industry gets disrupted the same way. It starts when only the rich can afford it and only experts can use/deliver it (e.g. telephones, air travel, computers) Then technology changes everything: - It lowers cost - Improves and standardizes quality - Makes it accessible to everyone A few examples of this throughout history: - NY to London call used to cost $300 for 3 min in 1930, and is now free. - Cross-country flight cost $4,400 in the 1940s, today it's $120. - A 1980 computer used to cost $10M, now you get your iPhone for $1K, which is 1000x more powerful. Healthcare is now on the same trajectory. - The cost of sequencing a human genome dropped from $100M to less than $100 in about 25 years - Full-body MRI used to be hospital-only, now you can get it for $500. - Blood panel went from $500 to $30. - CGM was prescription-only, now $70/month. The gates always fall.

I don't think people grasp how big this moment in healthcare is. - Diagnostics are cheaper than ever - More peptides are nearing approval - AI is starting to slowly outperform doctors All three curves are bending the right way at once. The future of health looks bright.

The cost of sequencing a human genome dropped from $100M to less than $100 in about 25 years. That's a million-fold decrease, which outpaces even Moore's Law. We're about to enter the era of personalized medicine.

YC wants health AI startups that: > Replace services (esp. healthcare admin) with AI end-to-end > Enable personalized medicine using genomics + diagnostics + AI agents > Build AI systems that make orgs (e.g. clinics) fully data-driven + self-improving Happy building! 🛠️

And here's how Claude Design rebuilt this blog → trivedigaurav.com/blog/how-claud…

Claude Design rebuilt and revived my 13-year old WordPress blog in minutes. First post on the new site? A Slack emoji generator that became team culture at Health Universe. Checkout: There's a meow for that! → trivedigaurav.com/blog/theres-a-… #AI #StartupLife #HealthcareAI

These ultra geniuses beat Sonnet 4.5 performance on LiveCodeBench with Qwen3-14B, a single RTX 5060 and a great harness.

It's finally happening. OpenClaw Mission Control just went multi-device. This changes everything about how you'll run your agents. Your M4 Mac Mini at home, your laptop at the coffee shop, your phone on the move-all commanding the same autonomous workforce. Open Beta drops very soon. Who's ready to architect their own AI empire? 🙋‍♂️

AI has brought back waiting for computer results.

Me Friday: "I'll just poke at OpenClaw for a bit" Me Sunday night: Making claws run on all Raspberry Pi sitting around The lobster has consumed me 🦞

After leading AI & Health at @UHN — Canada’s largest hospital network — for the past few years, this is the first time it genuinely feels different. A cardiologist just placed 3rd out of 13,000 at Anthropic’s hackathon. He built it between hospital shifts. In a week. An AI agent that follows patients home — reverse scribe, full history, devices, evidence — all in one place. A few years ago this would’ve taken a full team and months. Now? One doctor. That’s the shift.

AI doesn't need to be perfect to outperform our healthcare system's woeful status quo. It just needs to be better. In many cases, it already is. In today's @nytimes, I explore this idea – a central theme of my upcoming book, A Giant Leap (out Feb 3). nytimes.com/2026/01/19/opi…

On All Things Considered @npratc @NPR, I discuss patients' use of AI for health. While there are risks, many patients benefit. I prefer tools that guide patients through their symptoms (w/ iterative Q's & A's) rather than general tools like GPT or Gemini. npr.org/transcripts/nx…

Lots of (negative) AI references including AI scribes in the new The Pitt episode reddit.com/r/ThePittTVSho…

The full patient story, delivered in seconds. We've partnered with @Kno2 to bring nationwide record retrieval + AI clinical intelligence to health systems. Records from any connected provider → structured insights in minutes. linkedin.com/feed/update/ur…

Python 3.14 is out with the GIL (Global Interpreter Lock) removed. No more multiprocessing!