Ross Brodskiy is the director of the Computational Law Institute. He also founded Legawrite.AI, based in Aliso Viejo, California. The Institute publishes the research program he built around Legawrite.AI. His financial and proprietary interest in Legawrite.AI, and the rules the Institute follows because of it, are set out in Independence and Disclosure.
Research interests
His work asks what it would take for machine-assisted legal work to be filing-grade, and treats that as a question of legal theory as much as of engineering. Recurring themes:
- Determinacy. Separating research gaps from genuine doctrinal ambiguity, and measuring how far the outcome of a motion depends on who decides it.
- Usable law. Moving verification from whether a case is good law to whether a proposition is usable in this forum, at this procedural stage, for this client.
- Law as a versioned codebase. Governing enacted law, opinions, dissents and overruling the way software teams govern code, and designing interpretive canons that can be enacted before power is allocated.
- Design rules for legal AI. Precedence-ordered laws of system design, architectural grounding in place of after-the-fact verification, and open, rerunnable evaluation that publishes its failures.
- Adversarial reasoning. Opposition as a counter-model rather than a rebuttal.
- What remains for the judge. Decomposing judicial work into what a machine can verify and what remains discretionary.
He writes in English and in Russian.
Papers
- Good Law for What? A Proposition-Usability Model for Verification in AI-Assisted Legal Research. SSRN, 1 September 2026.
- isResponsive: An Open TAR Methodology for Court-Defensible Responsiveness Review. SSRN, 28 July 2026.
- Detecting Genuine Doctrinal Ambiguity: A Multi-Layer Framework for Identifying Structural Indeterminacy in Judicial Reasoning. SSRN, 23 July 2026.
- Ariadne's Thread: Legal Determinacy as Measured Dispersion across Posture-Instantiated Adjudicators, with Nathan Pokov. SSRN, 18 July 2026.
- The Promise Fulfilled: How Proposition-Level Legal Knowledge Representation at Corpus Scale Realizes Three Decades of Computational Jurisprudence. SSRN, 15 April 2026.
Essays
First published on Medium, now archived in the Library:
- The Verification Layer Is a Smoke Alarm, Not a Building Code, 20 September 2026.
- We Ran a Blind eDiscovery Review with No Humans in the Loop, 10 July 2026.
- From the Supreme Court Machine to Ariadne's Thread (in Russian), 21 June 2026.
- What Remains for the Judge When the Machine Has Already Verified the Law? (in Russian), 20 June 2026.
- I Built a Tiny Company Out of Robots, Gave Them House Rules, and Nearly Finished a Hard Thing, 15 June 2026.
- What a Best-in-Class Opposition System Must Actually Do, 12 June 2026.
- The Four Laws of System Design for Computational Law, 21 April 2026.
- Version Control for Law, 17 April 2026.
- Canons of Constitutional Interpretation of Planet Gibsonia, 17 April 2026.
- Case Treatment Classification Guide, 2 April 2026.
Elsewhere
- SSRN author page: papers.ssrn.com
- ORCID: 0009-0007-0913-3204
- Medium: @rossbrodskiy
Writing (15)
20 Sep 2026The Verification Layer Is a Smoke Alarm, Not a Building CodeWhy "just verify it" is the laziest idea in legal AI1 Sep 2026Good Law for What?A Proposition-Usability Model for Verification in AI-Assisted Legal ResearchRoss Brodskiy and Nathan Pokov · 6 min read28 Jul 2026isResponsiveAn Open TAR Methodology for Court-Defensible Responsiveness Review23 Jul 2026Detecting Genuine Doctrinal AmbiguityA Multi-Layer Framework for Identifying Structural Indeterminacy in Judicial Reasoning18 Jul 2026Ariadne's ThreadLegal Determinacy as Measured Dispersion across Posture-Instantiated AdjudicatorsRoss Brodskiy and Nathan Pokov · 8 min read10 Jul 2026We Ran a Blind eDiscovery Review with No Humans in the LoopOpen-sourcing a TAR engine, and publishing where it failed21 Jun 2026From the Supreme Court Machine to Ariadne's Thread20 Jun 2026What Remains for the Judge When the Machine Has Already Verified the Law?15 Jun 2026I Built a Tiny Company Out of Robots, Gave Them House Rules, and Nearly Finished a Hard ThingFive narrow agents, red lines, and the off-switch nobody designed12 Jun 2026What a Best-in-Class Opposition System Must Actually DoAI-Assisted Opposition Is a Counter-Model Builder, Not a Rebuttal Generator21 Apr 2026The Four Laws of System Design for Computational LawFour ordered design rules for legal AI17 Apr 2026Canons of Constitutional Interpretation of GibsoniaCanones Interpretationis Constitutionalis of Planet Gibsonia in the Guitaria System17 Apr 2026Version Control for LawA Git mental model for legal interpretation15 Apr 2026The Promise FulfilledHow Proposition-Level Legal Knowledge Representation at Corpus Scale Realizes Three Decades of Computational Jurisprudence2 Apr 2026Case Treatment Classification GuideCan I still rely on this case, and if so, to what extent?