AITomorrow's greatest founders. Featured today.
Unusual · Bet on himself when it wasn't the obvious choice. Now building the company that may define how brands survive the AI era.
Junior year of college. Goldman Sachs internship in hand. Return offer on the table.
The path was right there - the one that smart, high-achieving people follow without asking too many questions.
Then COVID hit. Keller Maloney asked for a gap year.
Goldman said if he took one, he wouldn't graduate on time, and they'd have to rescind the offer.
He took the gap year anyway.
"My options were take a gap year and turn down this offer, or suffer it out and work a job that could set me up for a career," he says. "I decided to bet on myself. If I really wanted it, I could try to get it again."
What happened instead was one of the most unconventional résumés in Silicon Valley - and eventually, a company built on one of the most contrarian theses in the current AI cycle.
Keller grew up in Los Angeles with a clear trajectory mapped out for him. His father was a doctor. His grandfather was a doctor. Medicine was the assumed next chapter.
What disrupted it was water polo.
He describes himself as a late bloomer - generally small, going up against a sport that is, in his words, "about as brutal and scrappy as it gets." He got obsessed anyway. He made the best high school water polo team in the country, possibly ever - playing alongside Ben Hallock, widely considered the greatest American water polo player of all time, and Olympian Johnny Hooper. For Keller, it wasn't the accolades that shaped him. It was what the sport required: relentlessness, physical grit, and the willingness to compete against people who were supposed to beat you.
That wiring followed him to Princeton, where he was recruited to play Division I water polo and became the youngest captain in the program's history. He came in planning to study physics. His first physics class had other ideas. He switched to economics and applied math, did internships at Canyon Partners and Ares, landed the Goldman Sachs offer, and was fully on the finance track his upbringing had pointed him toward.
Until he wasn't.
Berks County, Pennsylvania. Biden-Harris 2020 campaign. No politics background. No relevant experience. Moved his entire life there.
His job: convince strangers to volunteer their time. Deploy them to knock on doors and make calls. Build a grassroots organization from nothing.
He made 15,000 cold phone calls.
"It was super scrappy, super fun," he says. "One of the richest experiences of my life."
Coming out of the campaign, he crossed paths with Doug Schoen - one of the two inventors of overnight polling, a genuine innovation in political media. Keller, having never taken a politics class, ended up writing the manuscript for Schoen's most recent book, America Unite or Die.
Then solo travel. Then a chief of staff and head-of-growth role at Gatsby, an 8VC-backed startup - until the tech hiring freeze of 2022 swallowed the job before it started.
He sat down with Will, a mentor and friend five years his senior, and asked the obvious question: what now?
"He said, why don't we just tinker on something."
Five or six products later, three of which scaled past $100,000 in annual revenue, they landed on Unusual.
Jumping straight from college into building a startup is seen by many as risky. Keller argues it's actually economically rational - and he has a framework from his finance background to prove it.
He thinks about a person the way an investor thinks about an asset. When a stock is undervalued, you buy and hold it. When it's overvalued, you sell.
"That same thing actually applies to any person's career," he says. "A young person with high merit and little experience is an undervalued asset. The labor market prices that asset below its value."
The conclusion: bet on yourself hard early. Sell yourself hard later - that's when you consult, when you sell your time, when you let someone else hire you at full market rate. The worst place for someone with high merit is anywhere that won't let them rise to the level of it.
"Like the army - no matter if you are Achilles reborn, you are not going to rise to the rank of general before you're 50."
He describes his gap year résumé - 15,000 cold calls, a 300-person volunteer organization, a published book - as the most valuable twelve months he ever put on paper. Not by design. By doing what sounded interesting.
"I used to think it was just going to be a fun time. I'm not building my résumé in any way. Then I came back and it was by far the most valuable year for my résumé ever."
The best case scenario of betting on yourself in your 20s is remarkable. The worst case is just: try the next thing.
Will's background requires a moment to absorb.
He has been working in AI interpretability research since 2014. He built an early prototype of Starlink while at SpaceX. He's an MIT graduate. And at sixteen years old, he built a Farnsworth fusion reactor in his basement and gained international recognition for it.
"He's basically Tony Stark," Keller says, without irony. "He's probably the smartest person I know."
What makes Will rare isn't just technical depth - it's the combination of technical mastery and strategic imagination. He is the kind of person who, at sixteen, points his ambitions at nuclear fusion not because it's practical, but because it's the most interesting problem available. That orientation - toward huge, audacious goals - is as much what defines him as the engineering ability to pursue them.
The pairing is deliberate and complementary: Keller is the scrappy operator - the cold-call machine, the person who ran a political ground game across a state he'd never lived in, the former athlete who learned to outwork people twice his size. Will is the strategic and technical mind who has been studying how AI models think since before most people knew what a foundation model was.
Keller is clear that for him, the company almost didn't matter. "I would have done a hamburger stand with him. It was so much more about the team than any specific idea."
The idea they landed on is not a hamburger stand.
Most companies have no idea what AI models think and say about them.
That sentence is easy to skip past. It shouldn't be.
When someone asks ChatGPT or Claude which CRM to use, which bank to trust, which software to buy - the model answers. It recommends. It advises. It sometimes makes the decision entirely on the user's behalf. And the brands being discussed in those answers had zero input into how the model formed its view.
Unusual's thesis: AI models are not a new search engine. They are not a new marketing channel. They are a new audience - a thinking, evaluating, decision-making entity that brands now have to actively convince.
"It's a completely new persona to sell your message to," Keller explains. "You need to convince this intelligent, decision-making thing to love and recommend your brand rather than your competitor's. And that comes down to understanding how AI models think - which is fundamentally an AI research problem. Not a marketing problem. Not a sales problem."
This is the contrarian part. The current conversation in marketing circles frames AI as a new distribution surface - optimize your content for AI search the way you once optimized for Google. Keller thinks that framing is wrong and temporary. The real shift is that the audience itself has changed.
Unusual helps companies understand what models currently believe about their brand, why those beliefs exist, and how to reposition in a way that earns credibility with AI systems - not just humans.
Outbound first. Cold email as the primary channel.
And here's the detail worth paying attention to: Unusual uses its own platform as the top of funnel.
"Our platform tells us exactly what a company's core sales and marketing problems are and where AI models have misconceptions about them," Keller explains. "Which, interestingly, are actually very similar to what humans have misconceptions about them."
That becomes the personalization engine for every cold email they send. Pull a lead list from Apollo or LinkedIn, run it through an AI model to qualify against specific criteria, generate a targeted personalization based on something the prospect recently published or engaged with, drop it into an automated sequencing tool. Hundreds of emails a day.
The one constraint: the AI writes the template, but a human writes the copy. "LLMs write smelly copy right now," Keller says. "You can just tell. So we write a generally templated email and plug in a little bit of personalization. That combination converts."
Three people on GTM. Doing the work of 20 or 30. Keller's view is that adding a headcount right now would actually slow them down - the coordination overhead of human-to-human communication outweighs the throughput gain.
"It's the best time ever to be doing sales and marketing. And I'm not being hyperbolic."
They've raised $3.6 million in seed funding and are not in a hurry to raise again.
Keller is precise about the three things YC provides: network, signal, and capital. He's also precise about which one surprised him most.
"It's not even in our cold email copy. It's just that my LinkedIn says YC. And people take meetings because of it."
That signal, he argues, does more per dollar than almost anything else a founder can buy. Every investor in the world wants to see every YC batch company. The network connects you instantly to founders one stage ahead of your current problems. The philosophy is all published online - the real value of being in the room is watching that philosophy get applied across dozens of different companies' specific situations.
His YC partner was Dalton Caldwell.
"You get to do one rep of bringing a product to market and trying to sell it. And you feel empowered at the end that you have a process you can iterate toward. That was the real thing."
In the near term, one goal: prove that the product works. In a noisy space full of people talking about AI search and AI channels, Unusual wants to be the company that actually demonstrates results and reframes the narrative.
"AI models are not a new search engine. They are a new audience. We're the first company that helps brands market to agents as a new audience."
The 20-year version of that thesis is larger.
"AI agents are going to be making all the decisions for humans. Humans are going to be the finance department - approve, approve, denied, approve. And so the only audience that matters for a brand in 20 years is AI."
He says it without theater. Just a calm read on the direction of travel.
Every brand will need to convince a model. The model will need to be convinced in a fundamentally different way than a human does. And the team that has spent the most time understanding how those models actually think - not as a marketing problem, but as an AI research problem - will own that market.
Keller turned down Goldman because he believed he was an undervalued asset and the return on betting on himself was better than the return on being priced by someone else.
He was right then. He's building a company on the same logic now.
Keller Maloney is the co-founder of Unusual, a YC-backed AI research company helping brands understand and market to AI agents. He is based in San Francisco.
FoundersBrief - Tomorrow's greatest founders. Featured today