A research initiative of Nexlaw Partners

Applied AI for the business of law.

Real-world builds and evaluations, co-designed with firms and legal tech partners to put AI and automation to the test.

What's on the bench

Active projects

firmgen

An engine for generating synthetic firm data.

Automated methods and models that produce high-fidelity transactional data from the bottom up — driven by dozens of factors and compositional dimensions that define a firm's shape, size, and focus.

Screenshot of the firmgen experiment
interop

Own the AI integration layer.

MCP is the standard for connecting AI assistants to firm data. Tools-only servers aren't enough; renting it piecemeal from rotating vendors is a structural mistake. A brief on what MCP offers, why a firm-hosted gateway is the right shape, and what safe means in legal-specific terms.

Diagram: a firm-hosted gateway connecting AI assistants to firm systems
skillcheck

Evaluate AI skills using defined rubrics.

A bake-off harness for AI legal skills. Run controlled evaluations across models and prompts using researched “model-as-judge” techniques and severity-weighted rubrics.

Screenshot of the skillcheck experiment
harvest

Turn law firm websites into structured profiles.

An automated harvester that turns public firm websites into structured profiles — people, practices, offices, credentials — refreshed on a schedule and ready to feed downstream data and AI workflows.

Diagram: a law firm website extracted into a structured profile record
interviewer

Adaptive interview and insights platform.

An agentic interviewer and researcher that gathers information at scale through natural conversation — optimizing question structure against interview-defined data targets, previous answers, tone, and pace. An experiment in replacing rigid, tedious survey-based research.

Screenshot of the interviewer admin overview for a research study