Lucas Duarte Nunes Spent Nearly a Decade Building Automation. Then He Turned to AI Safety

By the time Lucas Duarte Nunes left Linte in 2025, he had spent nearly nine years helping build a São Paulo startup into a legal-technology company serving major enterprises across Latin America. He had led product, design, engineering, and later operations, negotiated important enterprise agreements, and helped guide the company to profitability. Leaving was not a rejection of the automation work that had defined his career. It reflected a growing interest in what happens when technology becomes powerful enough that efficiency is only one part of the problem.

"I still believe deeply in automation," Nunes said. "What changed was that I became much more interested in what happens when the technology becomes powerful enough that efficiency is no longer the only question."

His interest in automation began inside the legal profession. Nunes came from a modest family in Brazil and was the first in his family to enter the University of São Paulo. He chose law because he was interested in how technology should be regulated, but an internship at a law firm exposed him to another problem. Highly educated lawyers were spending large amounts of time drafting standard documents and handling repetitive tasks that software could perform more efficiently.

"At the time, it was obvious to me that a lot of this work did not need to be done the way it was being done," Nunes said. "I wanted to be on the side building the systems that removed that friction instead of preserving it."

In 2016, he left law school and joined Linte as an intern. Within months, his future co-founders invited him to become a co-founder, and together they began rebuilding the company. Nunes was elected a statutory officer in 2017. The following year, Linte raised a Series A from Redpoint eventures, Valor Capital Group, and 500 Startups. He went on to lead product, design, and engineering while gradually taking on a broader operational role.

His responsibilities eventually extended well beyond product development. Nunes negotiated and signed enterprise agreements, worked directly with major customers, and led implementations inside organizations using the platform. Linte's internal assessments recorded that replacing him would have required at least three hires, reflecting how many functions had accumulated under his role.

That breadth became one of the hardest parts of the job. Linte grew through Brazil's 2018 economic downturn and later through the COVID-19 pandemic, when the company shifted from years of in-person work to a fully remote model while continuing to sell to conservative enterprise buyers. Nunes was simultaneously closing contracts, running engineering, making product decisions, and taking on more of the company's operations.

"The founder's job can become wearing every hat at the same time," he said. "You learn a lot that way, but eventually the company has to become a system instead of depending on one person to hold everything together."

By 2025, Linte had reached positive EBITDA and was on solid footing. He decided to leave after nearly nine years with the company. The choice was difficult because so much of his professional life had been tied to building it, but he wanted his next chapter to connect more directly with questions about how increasingly capable artificial intelligence should be managed.

Nunes took a year-long sabbatical and traveled to San Francisco and London to meet people working on AI safety. His years building enterprise automation had made him optimistic about what software could eliminate, but they had also shown him how much trust and operational discipline were required before large organizations would depend on new technology. He began thinking more seriously about what those lessons meant as AI systems became more capable.

"I started out wanting to eliminate work that should never have required so much human time in the first place," Nunes said. "I still think that is worth doing. But as the systems become more capable, you also have to care about control and about what happens when they behave in ways nobody intended."

During his time in London, Nunes met the team at BlueDot Impact, an organization focused on building the workforce needed to address risks from transformative AI. He received a BlueDot Career Transition Grant and was invited to interview with the organization at the same time. Before the grant period ended, he had joined BlueDot as Head of Operations, where he now leads finance, people, and legal operations.

The move drew on skills he had already spent years developing. At Linte, Nunes had learned what it took to make automation dependable enough for banks, consulting firms, and other large organizations to use in real operating environments. That meant understanding how software interacted with existing processes, institutional caution, and the people responsible for keeping critical systems running.

At BlueDot, his role is different, but the operating instinct remains familiar. His responsibility is to make the organization around the mission work well, from finance and legal processes to the systems supporting staff and programs. Nunes sees strong operations as a way to protect the time of people whose expertise is better spent on the mission itself.

"The direction changed, but the underlying question is still familiar," Nunes said. "How do you automate what should be automated while keeping enough control and judgment around the things that matter most?"

For Nunes, that question connects the two phases of his career. The first was spent helping enterprises replace slow, repetitive work with software. The next is focused on ensuring that same trajectory, carried far enough, doesn't end with us losing control of the most powerful technology ever built.