Building an AI Attorney Matching Engine: Lessons from WOWERI

WOWERI LLC — Tue, 17 Mar 2026 17:52:45 +0000

At WOWERI, we've built an AI system that matches legal clients with the best-fit attorneys across 40+ countries. Here's how our matching engine works and what we've learned building it.

The Challenge

Legal matching is fundamentally different from other marketplace matching problems. You're not matching a rider with a driver — you're matching a person with a complex legal problem to an attorney with the exact right combination of jurisdiction, specialization, language, experience level, and pricing.

Get it wrong, and the consequences are severe — deportation, financial loss, or worse.

Our Approach

We process each client request through three layers:

Layer 1: Jurisdiction Filter — Automatically identifies which legal jurisdictions apply. This eliminates 90% of attorneys immediately.

Layer 2: Specialization Scoring — Each attorney has a specialization profile built from their case history and credentials. We score each remaining attorney against the client's specific needs.

Layer 3: Fit Optimization — The final layer considers language match, pricing compatibility, availability, and historical success rates to select the top 2-3 recommendations.

The Scale

488+ distinct legal service types
17 practice areas
163+ partner offices in 40+ countries
3 languages (English, Russian, Ukrainian)
100,000+ clients served since 2010
2,500+ cases closed

Tech Stack

React, TypeScript, and Supabase. Team of 65+ developers.

Key Lessons

Domain expertise > algorithm sophistication. Our best improvements came from working with practicing attorneys.
Quality control is non-negotiable. Every attorney is independently licensed and carries professional liability insurance.
Multi-jurisdictional legal tech is incredibly complex. Every country has different bar requirements and regulatory frameworks.