Work on Citational

Lawgame

Fri, 20 Feb 2026 00:00:00 +0000

Tl;dr: The AlphaGo for litigation.

What it is

Lawgame is an unsupervised AI system that simulates litigation through adversarial play. Three agents — Lead Counsel, Opposing Counsel, and Judicial Authority — war-game cases across multiple rounds to find dominant strategies.

The system doesn’t predict outcomes. It plays the game. Losses inform pivots. Wins get stress-tested. The goal is to find what human teams miss when they’re too close to the problem or too constrained by conventional thinking.

Shinshō-27B

Mon, 19 Jan 2026 00:00:00 +0000

Tl;dr: An Experimental Judicial Reasoning Model.

Shinshō-27B is my experimental quasi-reasoning model built on Gemma 3. Trained on approximately 20 million judicial opinions from the United States, United Kingdom, Canada, New Zealand, and Australia.

The aim is not another general-purpose legal language model. The aim is something narrower and more difficult: a model that captures how judges actually reason.

The Training Corpus

Twenty million opinions across five common-law jurisdictions. Appellate courts. Trial courts. Majority opinions, concurrences, dissents. The full texture of judicial decision-making, not a curated highlight reel.

MotionValidator

Mon, 19 Jan 2026 00:00:00 +0000

Tl;dr: A deep audit for court filings.

What this solves

A citation may be partly or entirely hallucinated. Maybe by AI, maybe by human error.

A citation can exist, have the correct pin-cite or neutral citation, and still be wrong. The case can be real, the quote accurate, and the application completely backwards. Or worse, the quote inaccurate, and the holding completely made-up. Surface-level citation checking misses the error that actually matters: whether the authority supports the proposition.

PriorStatement

Mon, 02 Mar 2026 00:00:00 +0000

Tl;dr: An assertion-comparison engine for litigation.

What it does

Most eDiscovery tools index keywords. PriorStatement indexes assertions.

It ingests the entire case file — transcripts, witness statements, police reports, medical notes, expert reports — and builds a profile of every assertion made by or attributed to each witness. Not just what they testified to in deposition, but what they told the police three months earlier. What the medical records say they reported. What third-party witnesses claim they said.