Natalie Aresta-Katz built Regbase, an AI-native legal research platform that tracks new and proposed laws across countries, languages, and jurisdictions, while earning her MBA at Columbia Business School. The startup was accepted into Y Combinator's Spring 2026 batch, where Aresta-Katz was one of just 18 women among the 431 founders, and closed an undisclosed seed round following Demo Day.
Regbase helps companies, law firms, researchers, and government vendors monitor regulatory changes that often appear on obscure government websites, consultation portals, and social media pages - sources that traditional search engines and large language models may never index. The company has already uncovered regulations posted only on LinkedIn, Facebook, or X, and identified 80 greenhouse-gas disclosure requirements across 45 countries for a global manufacturer.
From Big Law research to startup founder
Aresta-Katz developed the idea while working as an ESG researcher at Paul, Weiss, where she tracked environmental, social, and governance disclosure laws across more than 90 countries with a team of up to 13 paralegals. The work required running Google searches in 45 languages every month, and still missed regulations and produced human error.
"Me and a team of up to 13 paralegals were doing manual research about ESG disclosure laws all around the world," Aresta-Katz said. "It was extremely hard, even when we hired local lawyers, because everyone made mistakes and missed relevant laws. AI made mistakes and missed things too, but it was catching sources that nobody else could find, so I thought there might soon be a way to fully automate my job along with the attorneys overseeing the process."
She began building Regbase with co-founder Ethan Shea, a former staff engineer at Vercel, shortly before enrolling at Columbia Business School. A CBS professor, Mattan Griffel, who previously went through Y Combinator twice, pushed her to apply.
"I was talking to a CBS professor, Mattan Griffel, who previously did YC twice, about my company and he suggested I just submit an application and not overthink it," Aresta-Katz said. "Until then I had been bootstrapping and growing through revenue, so the world of venture capital was very new to me."
Building a company while finishing an MBA
Aresta-Katz folded Regbase into her MBA experience wherever possible, using it for class projects, slipping out of lectures for client meetings, and winning business from major law firms. She initially told the school she planned to drop out, but professors forgave her absences and the administration allowed her to take independent study so she could complete her degree.
"I think the truth at business school is that everyone has something else going on, especially in New York," she said. "I slipped out of classes to take meetings and tried to work my company into my actual course work. If we needed an example company to present on, we used Regbase."
Regbase addresses a real trust problem in legal AI. The platform maps every result back to the original government document, which Aresta-Katz says prevents the hallucinations that plague general-purpose AI tools. One proposed law the company found was hidden behind a QR code that required users to watch a 30-second ad on a mobile device before reading the document.
"Almost all hallucinations happen because some data or context is missing," she said. "Finding all of the sources is our specialty, and having the source documents prevents that issue."
The gender gap in AI startups
Aresta-Katz said she was surprised by the lack of women in her YC batch and in AI more broadly. She launched a Women of YC series on LinkedIn to feature each of her batchmates and encourage more women to apply.
"I didn't realize that the gender ratio at YC (and in AI more broadly) is actually getting worse, not better," she said. "During partner conversations I sometimes felt like I was speaking for every woman in the batch."
For legal professionals considering AI tools, Regbase's approach points to a practical lesson: the value of legal AI depends less on the model and more on the sources it can reach. For those working in legal roles, understanding how AI can track regulatory changes may become increasingly important - resources like this AI for Legal tag page and this AI Learning Path for Paralegals offer practical starting points for building those skills.
Why this matters for legal professionals
Regbase's early traction - including finding $400 million in expiring school-safety grants for a client selling protective equipment - shows what happens when legal research moves beyond keyword searches. For lawyers and compliance teams, the takeaway is concrete: regulatory monitoring that once required multilingual searches and local counsel coordination can now be automated, and the accuracy bar is set by whether the tool can trace every answer to a primary source. Firms that wait for clients to demand this capability, rather than adopting it proactively, will find themselves responding to requests instead of leading with answers.
Your membership also unlocks:
