Anteriormente Talisman AI
I went from building mobile games to solving problems for legal departments—and I’m having a blast.
I first met Mateus as an angel investor. Until then, I had no idea how the Brazilian legal system—a mix of good intentions and rampant abuse—creates a nightmare that cripples the economy. It became clear that this problem couldn’t be properly solved by humans in a cost-efficient way (emphasis on properly). Plus, computers seemed more than capable of doing a much better job at capturing all the details. And it felt like the tech was finally good enough.
AI has faced similar skepticism. Over the past 15 years, it introduced many new techniques but struggled to create an exponential change in our lives. Even after the rise of LLMs and generative AI, doubts remain: too much capital, too much energy, too much water, not real art, just copies, just hallucinations—you name it.
It’s easy to be a skeptic; it’s much harder to get out there and close the gaps.
Games didn’t need great graphics (though they got there eventually). Interfaces were adapted to touch screens. Multiplayer evolved with constraints until connections allowed for even fast-paced FPS. And games that could only exist in the massive, ubiquitous world of mobile were born.
Foundational model companies have done exceptional work. It’s insane how quickly models have become faster, better, and cheaper to run. The pace of change is so fast that today’s problem might not even exist—or might have a far better solution—next month.
Today, LLMs feel like talking to a weird genius with amnesia, trapped in a room with only a computer terminal for communication. They lack proper interfaces, memory, domain context, expressiveness, and sometimes intellectual honesty. Thousands of companies will need to rise to fill these gaps because nobody knows exactly how to do it. It’s reasonable to think these solutions will first emerge in verticals before we can finally set the genius free.
Did you know there are pumpkin throwing competitions with trebuchets in the US? While arguably a waste of time and resources, it highlights a proud culture of builders. Landing rockets is 90% willpower, 10% tech. Thanks to AI, many of today’s problems only need resilience to solve—and they’re just waiting for you.
I chose my horse and joined Mateus, along with my old friend Henrique from Wildlife. Many other builders have also left their comfort zones (and their skepticism) to create Enter, an enterprise service that helps companies handle challenges requiring exceptional accuracy and audit capabilities using AI—starting with mass lawsuits in Brazil.
Over the last year, we built our first product under a different name, Talisman. We designed it alongside great companies, partnered with Sequoia Capital, and expanded to serve over a dozen enterprise clients across multiple sectors.
