One-person startup: a Cornell postdoc wants to put AI in everyone's hands

I run my own AI startup, and I've built three AI products over the past eight months, from writing the code to landing the first paying users. I did it as a solo founder.

I came to AI from an unusual direction. I studied clinical medicine, then surgery, and could have become a doctor. During my PhD I turned to AI and stayed with it: I've published a fair number of papers at top AI conferences like ACL, CVPR, EMNLP, and AAAI, hold several patents, and did a postdoc at Cornell.

A record like that usually points to one path: stay in academia, aim for tenure. I chose to turn research into products, and put AI directly into people's hands.

Three products, one model I trained myself

All three products belong to the company I started, ModelsLive. The name is the whole idea: Models + Live, I want models to live in people's actual lives, out in the world. I trained the model myself, end to end, and all three products run on it.

Three products that look far apart are doing the same thing: taking the capabilities hardest to copy in humans and handing them to AI.

ProAdvisor gives AI social intelligence. Almost everyone has waited somewhere, full of questions, while the one expert stepped away. Empathy, conversation, reassurance, things we assumed only a person could do, turn out to be reproducible. ProAdvisor handles all of that before the expert steps in, so their time goes to what actually needs judgment.

Ekova gives AI the ability to reshape how you see yourself. People get stuck in a knot they can't name, sensing something is off while struggling to say what. Most AI companions make you feel better in the moment, which quietly deepens your dependence on them. Ekova works the other way: like a mirror, five personas help you see the patterns you keep getting caught in, so you understand yourself better over time and lean on it less.

VerbalValue gives AI an understanding of meaning. People buy who a product lets them become, as much as the product itself. A great salesperson can speak that meaning into someone's heart, a rare, hard-to-copy skill. VerbalValue hands that to AI and lets it do what a human finds hard: stay on all day, handle everyone at once, and remember every person who shows up.

Social intelligence, reshaping cognition, conveying meaning, the three capabilities hardest to copy in humans. I'm moving them onto AI, one by one. Since all three run on the same core, building three means one capability pointed at three problems.

I want to build AI that reaches everyone

So much of AI right now is aimed at people already inside tech, data platforms, agent frameworks, research tooling. Cutting-edge, and a world away from most people's lives.

I keep thinking about everyone else. Someone who has farmed for thirty years, run a shop for half their life, driven trucks for a decade, each of them is an expert in their own field. All of this has simply passed them by. Every one of my products works whether or not you know anything about AI. You open it when you need someone to turn to, when you want to get better at something, when you're trying to make a living.

To me, a technology's value is about how far it reaches, more than how high. Someone has to build the thing, but what makes it matter, over time, is all the people, in every field and every corner, who actually use it, in their own work, in their own way. They carry a technology forward. That's who I'm building for: to put AI into everyone's hands, and let the people who quietly hold up the world become the ones it truly belongs to.

It's early, real users, revenue just starting, plenty ahead. Happy to answer anything, about the build, the model, the solo grind, or any of it.