The bottleneck is data. The platform makes it useful in real time.
In a country where fewer than four in ten murders are ever solved, a team of engineers who left Silicon Valley to come home has built a platform that teaches fragmented police data to speak a common language. Pax, founded by David Peixoto and backed by $40 million from the investors who once saw Uber and Instagram before the world did, represents a quiet but consequential wager: that the gap between violence and justice in Brazil is not a fate to be accepted, but a data problem waiting to be solved. The largest seed round in Brazilian history is, at its core, a bet that technology can restore something cities lose slowly and notice only when it is gone — the feeling of safety.
- Brazil solves less than 40% of its homicides while the global average sits at 63% and Europe reaches 92%, a gap that costs the region $241 billion a year and keeps millions of people indoors after dark.
- Police departments across Brazil still cross-reference license plates through disconnected systems and review surveillance footage by hand — a manual bottleneck that Pax was built specifically to break.
- In Luziânia, the platform's first major deployment cut violent crime by 27% in six months, doubled police efficiency, and left 59% of residents reporting they felt safer — results that moved the needle far beyond the marginal.
- Greenoaks and Benchmark, the Silicon Valley funds behind Uber, Instagram, and Anthropic, have now placed their largest Brazilian bet on Pax, signaling that AI is migrating from consumer apps into the sectors that have resisted digitization the longest.
- With over 2,000 criminal cases resolved across 30 cities in a single year, the platform is crossing the threshold Andrew Cohen identified — from useful tool to something police officers can no longer imagine working without.
David Peixoto built his name in education technology, helping take Arco Educação public on the Nasdaq and cofounding an edtech company called isaac. Then he walked away from that world, returned to Brazil, and started asking a different question: what if the country's security crisis was, at its root, a data problem?
The answer became Pax — a platform that pulls together surveillance cameras, police records, license plate databases, and criminal histories into a single system capable of generating investigative leads in real time. Around Peixoto, a team assembled from Stanford, Harvard, MIT, and Brazil's own ITA and USP, many of them engineers who had left comfortable positions at major American tech firms to come back home and build something that mattered.
The results in the field have been striking. In Luziânia, a city in Goiás state, violent crime dropped 27 percent within six months of deployment. Police efficiency doubled. Nearly 60 percent of residents surveyed said they felt safer. Over the past year, forces using the platform have helped close more than 2,000 cases — homicides, armed robberies, vehicle thefts — across more than 30 Brazilian cities.
Now Pax has raised $40 million in what is officially Brazil's largest seed-stage funding round. The capital came from Greenoaks and Benchmark, the Silicon Valley investors who backed Uber before ride-sharing was a household concept and Instagram before its scale was imaginable. They also backed Anthropic. Their decision to lead this round reflects something larger than confidence in one startup: it signals that artificial intelligence is beginning to move into sectors — public safety chief among them — that have remained largely untouched by the technology wave.
The stakes are not abstract. Brazil resolves fewer than 40 percent of its homicides. The global average is 63 percent. Violence across Latin America costs the region roughly 3.5 percent of its GDP — $241 billion in 2025 alone. Peixoto's argument, and now his investors' argument, is that this is not an inevitable condition. It is a solvable one. The $40 million is a wager that they are right.
David Peixoto sits in a room full of data that most police departments have never learned to read together. Cameras feed into the same system as criminal records. License plates connect to stolen vehicle databases. In real time, patterns emerge that would take human investigators weeks to find. This is Pax, and it just raised $40 million in what amounts to the largest seed-stage investment Brazil has ever seen.
The money came from Greenoaks and Benchmark, two Silicon Valley funds with a track record of betting early on companies that reshape entire industries. These are the investors who backed Uber before it became synonymous with ride-sharing, who saw Instagram before anyone outside a small circle understood what it would become. They also backed Anthropic, the AI safety company. Now they are betting that an artificial intelligence platform designed to help Brazilian police solve crimes faster will matter just as much.
Peixoto built his reputation in education technology. He helped take Arco Educação public on the Nasdaq, then cofounded isaac, an edtech company. But he left that world to return to Brazil and build something different. Around him, he assembled engineers from Stanford, Harvard, MIT, and Brazil's own ITA and USP. Many had left comfortable positions at major American tech companies to come back home. The team understood something: Brazil's security crisis was not a problem that would solve itself, and it was not a problem that required accepting defeat.
The numbers tell the story. In Luziânia, a city in Goiás state, Pax ran its first large-scale operation. Within six months, violent crime fell by 27 percent. Police efficiency doubled. When researchers from Real Time Big Data surveyed residents, 59 percent reported feeling safer. These were not marginal improvements. They were the kind of shifts that change how people move through their own cities.
Over the past year, police forces using the platform have helped resolve more than 2,000 criminal cases across more than 30 Brazilian cities. The cases include homicides, armed robberies, and vehicle thefts—the crimes that make neighborhoods feel dangerous, that keep people indoors after dark. Each case represents an investigation that moved faster, a suspect identified sooner, a family that got answers.
But the scale of the problem remains enormous. Brazil solves fewer than 40 percent of its homicides. The global average is 63 percent. In Europe, it reaches 92 percent. The gap is not a mystery. Peixoto describes it plainly: the bottleneck is data. Police departments across Brazil still rely on manual work—officers analyzing video footage by hand, cross-referencing license plates through disconnected systems, searching through databases that do not talk to each other. The Pax platform takes that fragmented information and turns it into something coherent, something that can generate investigative leads and alerts in real time.
Andrew Cohen, a partner at Greenoaks, sees the opportunity clearly. "Ask Brazilians what problem they most want solved, and the answer is violence," he said. In the first state where Pax operates, vehicle thefts have plummeted. Police officers, he noted, can no longer imagine doing their work without the platform. That is the moment when a tool stops being optional and becomes essential.
The investment also reflects a broader shift. For decades, artificial intelligence has concentrated in consumer technology, in finance, in the kinds of sectors that attract venture capital easily. But AI is beginning to move into sectors that have remained largely undigitized—public safety among them. The violence in Latin America costs roughly 3.5 percent of the region's GDP, equivalent to $241 billion in 2025. That is not just a human tragedy. It is an economic fact. Peixoto believes the problem is solvable. He believes it is unacceptable to treat it as inevitable. The $40 million in funding suggests that serious investors agree.
Citas Notables
The bottleneck of investigation is a data problem. We built Pax from the ground up to connect real-world information and make it useful in real time. The police officer decides. The platform amplifies their capacity.— David Peixoto, founder and CEO of Pax
Ask Brazilians what problem they most want solved, and the answer is violence. Vehicle thefts have plummeted in the first state where Pax operates, and police officers can no longer imagine working without the platform.— Andrew Cohen, partner at Greenoaks
La Conversación del Hearth Otra perspectiva de la historia
Why does a seed round for a security platform matter more than, say, another consumer app getting funded?
Because it's solving a problem that affects how 200 million people live. When vehicle theft drops 27 percent in a city, that's not a feature improvement—that's a change in daily life. And seed rounds this large in Brazil almost never happen, which signals that major investors see this as a category-defining bet.
The platform connects cameras and databases. That raises obvious privacy questions. How does Pax navigate that?
The source doesn't address it directly, which is worth noting. The focus is on what the technology does—it integrates fragmented data—but the governance questions around surveillance and data use aren't discussed here. That's a real tension in the story.
Why did Peixoto leave a successful edtech company to build this?
He saw a gap. Education technology was working. But Brazil's violence problem was getting worse, and the tools available to police were stuck in the past. He had the skills to build something better, and he had the credibility to attract world-class engineers back to Brazil to do it.
The numbers from Luziânia sound almost too good. A 27 percent drop in violent crime in six months?
It's striking, yes. But it's also a pilot city—the conditions were probably ideal. The real test is whether those results hold as the platform scales to 30 cities, which is what's happening now. That's what the investors are betting on.
What does this funding say about where AI is heading?
It says AI is moving out of consumer tech and into sectors that have been stuck in analog workflows for decades. Police departments, hospitals, government agencies—these are the next frontier. And if it works in Brazil, it works almost anywhere.