CTech, the English-language tech publication of Israeli financial daily Calcalist, profiles Darrow co-founder and CTO Gila Hayat at the 2024 AI Conference — offering a rare window into the technical vision behind the platform from the engineer who built it.
Hayat opens with a reference that frames Darrow's mission immediately: Erin Brockovich, the legal clerk who built a landmark environmental case against Pacific Gas and Electric Company by going door to door in a small California town. The analogy is deliberate — Darrow is trying to do what Brockovich did, but at scale and with AI. The work involves recognizing the causality between real-world events and their legal significance, then turning that connection into legal action aimed at repairing damage done to the public. Consumer rights, environmental protection, privacy, and employee rights are the core domains.
The central challenge Hayat identifies is a double language barrier. On one side, the open databases that contain signals of legal harm are technically inaccessible to lawyers without programming skills. On the other side, advocates and journalists who find these stories often lack the legal fluency to assess whether they have a basis in law or precedent. These two groups — the technically skilled and the legally trained — rarely operate in the same space, which means potentially significant cases go unidentified. Darrow sits at the intersection, using language models to bridge both gaps simultaneously.
Hayat is careful to distinguish what language models can already do from what Darrow is building toward. Simple operations — document summaries, Q&A on legal text — are now accessible through off-the-shelf AI tools. The harder problem is complex, multi-step operations that require planning, strategy, and judgment about which stories have genuine public value worth pursuing. She cites the opioid crisis as an example of the kind of systemic harm that shaped careers, policies, and the relationship between the public and the justice system — the kind of issue that a well-designed legal intelligence system should be able to surface before it reaches that scale.
On the question of whether AI will replace lawyers, Hayat pushes back on the early panic that followed ChatGPT's release. The profession is developing a more mature understanding of AI's role in knowledge-intensive fields, she argues. The direction of travel is toward an agent-based model, where humans serve as action planners and AI handles the information infrastructure — a considerably more sophisticated framing than the document summarization use cases that dominate most legal AI discussion.
At the time of publication, Darrow employed approximately 120 people across Israel and the United States, and was actively recruiting lawyers and software engineers to join its mission.
