Most health-AI conversations are about what to build. This one is mostly about what not to. Smit Doshi spent a decade leading global IT before helping build MyHappy.Health — a free, voice-first AI health companion that works in more than two dozen languages, made for people who often have never seen a doctor and have no second opinion to fall back on.

That constraint sits at the center of the conversation: when your user has no other option, a confident wrong answer isn't a bug, it's a danger. Smit walks through why his team built a working anemia screener and then left it deliberately switched off, why the assistant greets people as Sam or Maya instead of a chat box, why they refuse to collect user data even though it would make the product better, and why the feel-good promise that "AI will give everyone a doctor" is the story he'd push back on hardest.

"A confident wrong answer or a hallucination from AI can harm or kill someone. And that's the risk that I don't wanna take — not just for myself, but for humankind." -Smit Doshi

Connect with Smit Doshi

LinkedIn

My Happy Health

Hosted by Kelly Smith. Find more at kellysmith.me/healthcareux.