He coaches couples through breakups by day and debates the extinction of humanity by night. The throughline is an old-fashioned faith in arguing things out.
Blue sky behind him. P(doom) in front of him.
In 2012, Liron Shapira put $10,000 into a small exchange called Coinbase. Nine years later he sold for roughly six million dollars. Most people would frame that as the story of their life. Shapira uses it as Exhibit A in his argument that cryptocurrency is, in his words, essentially worthless. He made the bet, took the money, and kept the opinion. That combination - profiting from a thing while refusing to flatter it - is about as close to a personal philosophy as he gets.
Today he runs two operations that look, on paper, like they belong to different people. One is Relationship Hero, a Y Combinator-backed coaching service he cofounded and leads as CEO, which has walked more than 100,000 people through the hardest conversations of their lives: the breakup text, the reconciliation, the should-I-stay. The other is Doom Debates, a podcast and YouTube show he built to drag a single question into the mainstream - whether artificial superintelligence is about to kill us all, and what, if anything, we can do about it.
"Growing this show's subscribers actually has significant leverage to lower P(doom)."
Relationship Hero is the quieter half of his life, and in some ways the more telling one. Launched through Y Combinator's Summer 2017 batch, it offers on-demand coaching for people in the thick of romantic decisions - not therapy exactly, but a trained voice on the other end when the stakes feel impossibly high. The company has grown to a team of dozens and a client roster past six figures. It's a profitable, durable business built on an unglamorous truth: people will pay for a clear head in a moment of chaos. That's the same product, more or less, that he's now trying to sell to the entire AI discourse.
The bridge between coaching couples and debating extinction is not as wide as it looks. Both are bets that humans can think their way out of trouble if someone forces the conversation to stay honest. Shapira is a creature of the online rationalist community - the loose movement built around clear reasoning, calibrated probabilities, and the unfashionable idea that you should say what you actually believe. He assigns roughly a 50% chance that AI causes human extinction by 2050. He does not say this to scare you. He says it because that's his number, and hiding it would be a kind of lie.
Before any of this, there was Quixey. Shapira cofounded it in 2009 and served as CTO, inventing what the company called Functional Search - a way to find apps by what they do rather than what they're named. Quixey raised over $170 million, including strategic capital from Alibaba, and became one of Silicon Valley's brightest app-era hopefuls. Then it collapsed. Where most founders would reach for a softer word, Shapira reaches for the hardest one available: a massive failure that destroyed far more value than it created. The honesty is the point. It usually is.
He grew up into computer science at UC Berkeley, the kind of foundation that lets a person treat a love life, a startup, and the fate of the species as variations on the same problem: define the terms, weigh the evidence, and don't flinch from the answer. Watch enough of his work and a pattern emerges. He is drawn to the moments where everyone else has agreed to stop arguing - crypto is the future, AI is fine, the breakup is final - and he plants himself there and reopens the case.
That instinct made him a fixture of tech Twitter long before Doom Debates, where he built a reputation writing threads that picked web3 apart use case by use case. His definition was precise and inconvenient: web3, he argued, is blockchain applications other than cryptocurrency, and most of them solve problems that don't exist. It won him critics and a following in roughly equal measure, which is the natural yield of saying things plainly.
What he's chasing now is bigger than a company. Doom Debates is an attempt to build social infrastructure - a durable, watchable format where the smartest optimists and the most worried pessimists actually engage, on camera, until something is resolved. He brings on leading AI researchers. He lets them make their best case. Then he tests it. The format assumes the audience is intelligent and the stakes are real, and it refuses to pretend otherwise.
Consider the through-line of his arguments. With crypto, he watched a consensus form - this is the future of money, get in early - and asked the unwelcome question: what does it actually do that existing tools don't? With web3, he extended the same scalpel, separating the cryptocurrency from the cathedral of applications built on top of it and finding most of the latter hollow. With AI, he's run the move in reverse. Here the consensus is comfort - the systems are useful, the risks are overblown, the smart people have it handled - and he plants himself against it, insisting the danger is real and the complacency is the problem. The topics change. The method doesn't.
That method has a name and a home. Shapira came up through the rationalist community, the milieu shaped by writers like Eliezer Yudkowsky and institutions like the Machine Intelligence Research Institute, where Shapira serves as an advisor. He also advises the Center for Applied Rationality, an organization built around teaching people to think more clearly under uncertainty. These aren't ceremonial titles. They're the intellectual scaffolding behind everything he makes - the reason he talks in probabilities, the reason he prizes calibration over confidence, the reason a debate format felt to him like the natural unit of progress.
Doom Debates is the clearest expression of that worldview. The premise is almost defiantly simple: bring the most serious AI optimists and the most serious worriers into the same room, give each their best shot, and don't let either side hide behind vibes. He interviews leading researchers, walks through their reasoning step by step, and presses on the cruxes - the specific points where reasonable people diverge. The bet is that the bottleneck on AI safety isn't only technical. It's social. If the public conversation were sharper, more people would understand the stakes, and understanding, he argues, has leverage.
It would be easy to file Shapira under doom-monger and move on. That misreads him. The through-line of his career is not pessimism. It's a stubborn optimism about argument itself - that a clear enough conversation, held in public, can move a number that matters. He turned an early Coinbase bet into a fortune and an opinion. He turned a failed search engine into hard-won candor. And he's turning a probability most people would rather not think about into a show you can subscribe to. The wager underneath all of it is the same one he's been making since Berkeley: that thinking out loud, carefully, is leverage.
Most people in tech keep their real estimate of AI risk politely vague. Shapira refuses the courtesy. His P(doom) - the probability he assigns to AI causing human extinction - sits around a coin flip by 2050. It isn't a marketing figure. It's a calibrated belief, stated plainly, which is the entire ethos of the rationalist world he came up in.
The point of Doom Debates isn't to make you despair. It's to make the conversation good enough to actually move the number.
Becomes CTO of "The Search Engine for Apps" and invents Functional Search - finding apps by what they do, not what they're called.
Makes an early angel investment in a young exchange named Coinbase.
After raising $170M+ including Alibaba capital, the company shuts down. Shapira calls it a failure that destroyed more value than it created.
Cofounds the coaching startup and ships it through Y Combinator's Summer 2017 batch.
Sells his Coinbase stake for roughly $6M - and keeps arguing that crypto is essentially worthless.
Builds a podcast and YouTube show to drag AI extinction risk into the mainstream and host high-quality debate.
The app-store search engine. Big vision, $170M+ raised, Alibaba on the cap table - and a collapse Shapira describes with unusual bluntness.
On-demand coaching for the messiest moments in love. 100,000+ clients, a YC pedigree, and a steady business built on hard conversations.
A show built to argue about AI extinction in public, with the field's best minds, until something actually gets resolved.
Growing this show's subscribers actually has significant leverage to lower P(doom).
Raise mainstream awareness of imminent extinction from AGI. Build social infrastructure for high-quality debate.
Quixey destroyed way more value than we created.
He turned a $10K Coinbase bet into roughly $6M, then spent years arguing the whole asset class is worthless. He sees no contradiction.
A personal one (@liron00) and Doom Debates (@DoomDebates) - one for him, one for the end of the world.
His personal blog carries a name that doubles as a critique of startup excess.
He advises both MIRI and the Center for Applied Rationality - two load-bearing institutions of the movement.