Emily Cabrera, a University of Miami School of Law alumna and legal counsel at IRG Ventures LLC, won the April 2026 HeyCounsel Claude Cowork Hackathon with a tool that stops lawyers from accidentally waiving attorney-client privilege before they even hit send. Her project, Privilege Sentinel, acts as a pre-flight checker that pauses a lawyer when a draft prompt risks disclosing privileged information-a vulnerability she said most AI risk discussions overlook.
From weekend hackathon to global validation
The five-hour virtual event brought together hundreds of lawyers from over 10 countries to build open-source AI tools using Anthropic's Claude Cowork and Claude Code. Cabrera had been working with AI for only two months and had no formal computer science background. Her winning entry was quickly integrated into open-source legal skills ecosystems like Lawvable and LegalQuants. When Anthropic later highlighted those community hubs during its "Claude for the Legal Industry" rollout, Privilege Sentinel gained global visibility.
"The dangerous moment is before the prompt leaves the lawyer's machine," Cabrera said. "Privilege Sentinel became very obvious to me. It had to be a pre-flight checker... something that helps a lawyer pause before the risky disclosure happens."
Legal intuition as a design asset
Cabrera credits her approach to experiences at Miami Law, particularly the Law Without Walls program with Professor Michele DeStefano and an intellectual property course with Professor Andres Sawicki. "Law Without Walls…was the first time I understood that law could be a design discipline," she said. "Look at a problem in the world, assemble an unconventional team, and build a better way to solve it." Sawicki's class taught her to ask, "Who owns the value, and why?"-a question she still uses with AI-enabled companies.
She said the transition to building with AI felt more like drafting a legal document than learning to code. "You need precision. You need definitions. You need to anticipate edge cases. Lawyers are already trained for that. I just had to learn to aim that training at a different kind of document."
What she tells law students
Cabrera urges current law students to start building small tools before they feel ready. "Pick a legal problem you actually care about. Not 'AI and law' in the abstract-something narrow and irritating and real," she said. "Build the smallest possible tool that improves that one thing." She is already applying that advice to her next open-source project, Counterweight, which analyzes contract risk through bargaining-power asymmetry.
Why this matters for legal professionals
Cabrera's win signals a shift in who builds legal AI. The tools that will shape practice are increasingly created by lawyers who understand both doctrine and system design, not just by software developers. For legal professionals, the takeaway is concrete: hands-on familiarity with AI tools like Claude is becoming as practical as knowing how to structure a contract. Courses that teach AI for Legal Professionals can help bridge the gap between legal expertise and technical execution, while Claude AI Courses offer direct experience with the platform Cabrera used.
"The next generation of legal tools should not be built around lawyers," Cabrera said. "Lawyers should be building them."
Your membership also unlocks:
