## TL;DR
AI legal research finds the needle in the haystack -- semantic search across millions of cases, predicting judge rulings, and generating legal memos. From CoCounsel (first AI to pass the Bar) to Westlaw Precision, AI is becoming the junior associate that legal research teams rely on.

## Core Explanation
Legal research AI: (1) Semantic search -- embedding-based retrieval (LegalBERT, CaseLaw-BERT) finds conceptually similar cases beyond keyword overlap. "Negligent infliction of emotional distress" finds cases using different terminology; (2) Citation analysis -- Shepardizing (tracking treatment history: affirmed, reversed, distinguished, overruled). AI predicts citation trajectory; (3) Litigation analytics -- predict: motion outcome probability, case duration, judge tendencies, damages range; (4) Generative -- LLM-drafted legal memos from natural language research questions.

## Detailed Analysis
CoCounsel (Casetext, 2023-2025): GPT-4 powered. Features: legal research memo, document review, contract analysis, deposition preparation. Passed the Uniform Bar Exam (first AI). Acquired by Thomson Reuters ($650M, 2023). Westlaw Precision: AI-enhanced search with "issue-specific relevance" ranking. Lexis+ AI: conversational search, AI-generated case summaries, Shepard's integration. Lex Machina: predicts case outcomes and judge behavior. Database: 5M+ federal cases, 100K+ state cases. Judge analytics: motion grant rate, time-to-decision, appeal history. Law firms use for venue selection (which judge/court is favorable) and settlement valuation. Key challenge: hallucination -- AI-generated legal content must be factually verified. CoCounsel provides cited sources linking every claim. Competitor attacks: ROSS Intelligence shut down (2020) due to Thomson Reuters copyright lawsuit over training data.