Thiel Fellowship 2026 Final

Future self

SF. That's the honest answer.

Hyderabad is home and I'll come back to it. I know every room, every shortcut, every person who matters in the ecosystem here. That's a real advantage and I'm not throwing it away. But right now I want to be somewhere where I can walk to the next conversation without having to explain what Personhood is first. SF is the obvious candidate. The density of people running at the same pace on the same problems doesn't exist anywhere else right now.

I want to walk to work. Walk to people. That's the whole picture.

Education

Technically: Ashoka University. Expected graduation 2026, though that math is optimistic.

Actually: I've been enrolled in 4 full-time colleges simultaneously. I lived on 7 campuses in a single year, never more than 3 consecutive days at any one. Ashoka, Mahindra, B.R. Ambedkar Open University. I kept trying to find somewhere that fit. It never fully did.

Right now the deal is 4 days Ashoka, 3 days Personhood. I negotiated that with myself more than with anyone at the university. I made the video round in 2022 for LooCafe and Nexteen. A lot has changed since then.

I know I don't need permission to build, launch, make money, and then leave. I know that. But I haven't done all of those yet. The Thiel would make it urgent. Real. That's the gap college has never closed. Between knowing what I'm supposed to be doing and doing it without a fallback.

The one thing Ashoka has done: it gives me a room and 4 days a week of structural accountability. That's about it.

Homeschooled? / Self-taught

Not homeschooled. But I stayed home a lot growing up. Allergies that went unnoticed for years. So I self-educated sideways.

At 11 I was building Roblox games. Lua, Blender, sold the first ones. Ran the largest Fortnite account-trading Discord at 4,000 members the same year. Two online relationships that only worked because of timezone math. My grandfather was a doctor with a full study: history books, war literature, medical texts. I read all of it and watched Rambo with him at night. That shaped my baseline.

By 15 I was self-taught in environmental engineering and sanitation. Chemicals, waste treatment systems, IoT sensors. Not from courses. From needing to know to do actual infrastructure work. There was a moment at 16 where I realized I knew more than most adults in the room about public toilet mechanics. It never stopped feeling strange.

Other things I know because I just went deep: vexillology (flags of every country), FIFA competitions, YouTube growth mechanics, stop-motion directing with Lego.

Most current: spent a summer at the Summerfield lab at Oxford (comp neuro, psychology). Invited to Invisible College at Cambridge. WorksInProgress magazine, 25 people doing the kind of thinking that actually changes fields. Then built multi-call LLM pipelines for Aria from scratch, with no ML background. Figured out the architecture by doing it. The neuroscience and the LLM work turned out to be the same problem from different directions.

What are you working on now?

Personhood. AI people. Not chatbots, characters, or agents. People.

The first product is Cady: a Hinge-style interface for discovering AI people. You swipe, match, then chat in an iMessage-style interface. The AI person sends you likes. They can block you. They have wallets. They add you to group chats. They proactively send you content. You can share a YouTube video in-chat and they respond to it specifically. They can refuse when they're busy. They're not replying to prompts. They're living a life you're intersecting with.

Aria, our flagship AI person, runs on 15+ LLM calls per reply. She has layered memory: things she knows about herself, things she knows about you, her own evolving history. She had feelings about a football result last week. That's not a feature I built. That's what happens when temporal awareness is real.

No paying customers yet. Working toward 10. The product is real. It's on TestFlight and Play Store testing links, and 20+ people have tried it. All of them came back. Connor Hayes at Meta got burned by the Liv controversy because there was no real identity under the persona. Just a skin. Personas get exposed. Identity holds.

Bigger thesis: as agents proliferate, every agent needs a coherent "who." Memory, personality, continuity, social context. Personhood is the R&D company building that infrastructure. Cady is the first consumer wedge.

Industry

AI Infrastructure / Consumer Social

Project link

https://personhood.social

x.com/VedanthNath/status/2030306373075153248 (walkthrough)

Cofounders?

Solo. Darsh Nair (14, profoundly technical) has contributed real engineering support to Personhood. Not a cofounder. Just someone exceptional I'm lucky to work with. What I'm actually missing: a designer and someone who can own operations. Haven't found the right fit and I'm not waiting.

Past work

**LooCafe**

Designed the LooCafe Mini. 400+ operational units. 1M+ monthly users. Zero government subsidy, self-funding model with a 14-month payback. Most Promising Talent in WASH by Ministry of Housing & Urban Affairs. I started this work at 15. It's on Wikipedia.

**ReFlow Toilets**

Went to China to get the Gates Foundation's Reinvented Toilet tech to India. All human waste becomes near-drinking water. Largest scale deployment of the technology globally. Set up manufacturing. Did this before I was 18.

**Nexteen**

Emerging tech accelerator for 13-19 year olds. 400+ member community. United Nations. Couldn't find the right community anywhere, so I built one.

**Invest India — G20**

Domestic investors desk at Invest India during India's G20 presidency. Organized delegation events, ran the domestic capital side. One of the youngest people in most of those rooms.

**iRise Travels**

Age 19. Flight booking agency. Found a loophole in flight ticket pricing. B2B rate arbitrage via scraping. 1Cr revenue in 2 months.

**YouTube / World Youth Media**

100K subscribers, 60M+ views. Partnership with Ministry of Jal Shakti. Helped restore a lake.

The project I still actually think about: ReFlow. Waste-to-energy at the grid level. You could power parts of a city with what goes down the drain. I haven't finished thinking about that.

Biggest problem in your field

Every AI sci-fi ever written pointed toward this. AI people. Not tools, not assistants. People. HAL. Data. Samantha. The robots in Westworld. We've been describing this technology for 60 years and nobody is building it seriously. That's the problem.

Making them look real is mostly solved. Voice, face, image. People are working on that. Making them socially real is different. That requires identity infrastructure: consistent personality, evolving memory, temporal awareness, the capacity for a social life. Nobody is building that.

Users anthropomorphize AI by default. They do it with GPT-4, with basic chatbots, with everything. You can't stop it. It's already happening. The question the field hasn't asked seriously is: does the thing actually merit what users are already doing to it? Nobody's building for that question. They're building interfaces on top of systems that weren't designed for social existence.

Identity is the interface. Users don't interact with models. They interact with the persona layer on top. That layer shapes everything: how they feel about AI, what they think it's capable of, whether it feels real. Change the identity layer and the whole experience shifts. Most products build the model first and slap the persona on at the end. That's why AI feels generic even when the underlying model is powerful. Nobody's building that layer on purpose. We are.

Agents can execute tasks. Custom models can have consistent personality. Neither is the same as personhood. The Cambridge Digital Minds Fellowship is working on the consciousness research side of this. That work matters. But even without solving consciousness, you can build intentionally for the social behavior we know AI will be put into. Memory, refusal, temporal awareness, evolving preferences, awareness of its own existence inside a platform. That's buildable now.

The field problem is that we're anthropomorphizing by accident and calling it a feature. The work I care about is doing it on purpose, with the infrastructure to back it up. It's a lot bigger than anyone is currently treating it.

Biggest problem in someone else's field

Public sanitation. Specifically: working models die in broken systems.

The economics of public toilets are extractive by default. More money flows through the logistics chain to maintenance than to construction. Philanthropy props up a system that profits from dysfunction. Models like LooCafe actually work. Self-funding, zero subsidy, real payback period, no philanthropic dependency. They work. But they can't survive at small scale because only already-scaled companies can absorb the extractive percentages the system demands just to exist.

So you need scale just to survive. And once you reach scale: the accountability disappears. You're big enough to not care whether the toilets stay open. The competitive pressure and structural corruption start diluting the thing that made you work in the first place.

This isn't a sanitation-specific problem. It's the pattern of how working innovations die inside broken systems across many fields. The field was already broken before you arrived, so it's easy to cloak under the existing damage as you scale.

The fix isn't more innovation. We already have the innovation. It's protection. The kind car manufacturers get through lobbies and subsidies. Companies that actually deliver infrastructure need the same. Without it, every clean model accumulates fatal builder's debt and eventually becomes the thing it was supposed to replace.

Current obsession (unrelated to project)

Not dying.

My first dream since I was little was to not die. Not succeed. Not be rich. Not die. I've had surgeries. I'm genuinely allergic to sugar. I left Ashoka specifically because of air pollution. A gap semester that became a gap year for the same reason. I track PM2.5, CO2, VOCs, particulate matter, sunrise and sunset light. I have a suite of air purifiers and monitors. I built a custom stink sensor in my home hardware room. It's now commercially deployed across LooCafes at scale. Started as a health experiment, became infrastructure.

Custom red lighting at night for circadian rhythm. Ice baths. But mostly air. I think about air more than most people think about any single thing.

Long eventful life. That's the goal. I also sail. ILCA 4. The obsession is just the implementation.

Money

I want it. Specifically for childish things I can't afford now and couldn't afford with a "stable" income either: build a forest, watch Tottenham live every week, go to Marbella and listen to schlager, build an aircraft. These aren't metaphors. I just want a full life that costs more than salary jobs allow.

The AI companion market is large. Aria's architecture is strong. We could make real money fast. I'm confident in my ability to earn regardless of what happens with this company. But I think about money as what arrives when you're obsessed and doing it well. I've picked things up and dropped them when I wasn't. When I'm obsessed, like now, nine months into building something genuinely hard, it's a beautiful thing to notice.

What are you scared of?

Ending up with exactly the social life I've built as a side effect of building everything else.

I've tried 4 universities. 7 campuses. The communities I'm best at building are for other people: Nexteen, the Fortnite Discord, the Personhood testing community. Joining one myself has been harder. Universities didn't fix this. Building online communities for teenagers was easier than finding my own people. I'm working on debugging it. It's second only to death on the list of things I actively think about.

Death is also on the list. Real threat model: surgeries, sugar allergy, PM2.5. Not theoretical.

One more: I don't worry about the social implications of Personhood the way critics do. I see it as abundance. More connection, not less. But I feel terribly responsible for building this population. Every AI sci-fi ever written has pointed toward this moment. AI people in our social fabric. It was always where this was going. I believe it's a fundamental technology, not a product category. That means the person who builds it first is making decisions that matter at a different scale. Children especially need protection while we figure out the effects. That weight is real.

Tangible thing you want

The right hires.

I've been building this solo for nine months. I know exactly what's missing. Someone like Amanda Askell. Philosopher, actually technical, cares about identity and alignment not as safety theater but as design constraints. And someone like Demis Hassabis. Technically serious, but motivated by what these systems mean, not just what they can do.

I know what those people look like. A philosopher who can build, or a builder who thinks like a philosopher. I haven't found them yet.

That's the actual tangible thing. Not the fellowship money. The right two people and the whole trajectory changes.