Twenty-four hours without sleep, a Capital One hackathon at Talent Land and a finished product: InversoAI. We made the top three of the Genius Arena. Here's what we built and why the judges liked it.
The challenge
The Genius Arena Hackathon, presented by Capital One inside Talent Land 2025 in Guadalajara, posed a problem you feel every day in Mexico: people don't know finance, and the material that exists to learn it doesn't speak to anyone in particular.
Personal finance courses are all the same. They explain compound interest with the same example for the 19-year-old student, for the woman who runs a corner shop and for the 40-year-old engineer planning retirement. And because they speak to no one, nobody finishes them.
We had 24 hours. We didn't sleep at all.
What we built: InversoAI
The bet was simple to state and hard to execute: a finance tutor that gets to know you first and teaches you second.
The flow has four steps:
1. It measures what you know
Before teaching anything, the system asks questions we trained to locate the person's actual level. Not an exam: a conversation. Nobody feels tested, but by the end we know whether we're talking to someone who has never heard of treasury bonds or someone who already has a portfolio.
2. It figures out who you are
Woven into those same questions is what really matters: what you like, what music you listen to, which people you follow, what tone you connect with. It sounds irrelevant for a finance app. It's the opposite: it's what makes the next step work.
3. It builds a program made for you
With those two inputs —level and profile— it generates a personalized finance program. Not a generic course with your name on top: the explanations, examples and analogies come out of what already interests you. If you like football, risk gets explained with football. If you follow a certain creator, the tone resembles theirs.
4. It takes you to real instruments
And this is where it stops being a course. Based on your program and your investor profile, the system connects to catalogs of real instruments: from government bonds to mutual funds and more advanced options. The bridge between "I get it" and "I invested" is where almost all financial education falls apart, and it was the one we wanted to fix.
Why we made the final
Three things, I think.
The personalization was real, not a promise. The demo showed the same financial concept explained two different ways for two different profiles, generated live. That lands without anyone explaining it.
It closed the loop. Many financial-education projects end at learning. Ours ended at a concrete instrument the person could actually buy.
It was finished. In 24 hours we shipped something usable, not a mockup. Two repositories, the assistant and the platform, working together.
What I took away
That AI personalization isn't cosmetic when the subject is hard. In finance the barrier is almost never the person's intelligence: it's that the material doesn't connect with their life. A model that knows who you are can close that gap, and that changes how many people reach the end of the road.
Something about the format became clear too. A 24-hour hackathon doesn't reward the biggest idea: it rewards the clearest idea that can also be shown running.
Frequently asked questions
What is the Genius Arena Hackathon?
It's the hackathon presented by Capital One within Talent Land México 2025, held in Guadalajara in April 2025. Teams build a project in 24 hours and present it to a panel of judges.
Where did John Olven place?
His team made the top three finalists of the hackathon with InversoAI.
What is InversoAI?
A personalized financial-education platform powered by AI. It assesses the user's knowledge level and interests through conversation, generates a finance program tailored to them and connects it to real investment instruments, from government bonds to mutual funds.
Who is it for?
Young adults in Latin America who want to learn about finance and make their first investment, and who don't stick with traditional courses.