Legal eDiscovery

What We're Exploring

Re-architecting discovery so relevance prioritization, retrieval, and review guidance happen continuously at ingestion—reducing layered human-only review cycles while preserving auditability, defensibility, and precision.

The Problem and Why Now

High-stakes litigation discovery can cost $10M+ and run for months because teams manually review enormous, heterogeneous document sets in sequential layers. Those timelines don’t just increase cost—they shape strategy, settlement dynamics, and leverage between parties.

Courts are increasingly accepting AI-assisted review as part of defensible workflows, while language models can now operate with higher precision across complex corpora. With the right guardrails, the discovery bottleneck can be structurally compressed.

Our Hypothesis

If discovery can surface the most relevant materials early with auditable reasoning, then timelines collapse, costs drop structurally, and legal teams regain strategic speed without sacrificing rigor.

What Must Be True

That courts and opposing counsel will accept AI-driven prioritization as a default primitive once timelines compress materially—and that the system can consistently meet defensibility and accuracy standards across matter types.

Who We're Learning From

General counsel, litigation partners, and discovery leaders managing high-volume, high-stakes matters.

Focus Geography

United States (initial focus: New York / San Francisco)