The AI Future of Law Is Already Here — It’s Just Not Evenly Distributed

Why It Matters

AI can deliver dramatic productivity gains for contract‑focused firms, reshaping competitive dynamics, yet many practice areas face privacy, accuracy, and workflow barriers that limit adoption.

Key Takeaways

• Off‑the‑shelf models outperform niche legal AI tools
• Claude’s “skills” automate contract drafting steps
• Prompts average 2,000 words for high‑quality output
• Efficiency gains concentrated in transactional, startup practice
• Litigation faces privacy, hallucination, and accuracy challenges

Pulse Analysis

The legal industry is at a crossroads as general‑purpose large language models (LLMs) eclipse niche AI platforms built for lawyers. Shapiro’s experience shows that practitioners who abandon wrappers like Westlaw’s Harvey and engage directly with Claude or ChatGPT can tap into raw model capabilities, cutting development time and cost. This shift mirrors broader enterprise trends where companies favor flexible, API‑driven AI over point solutions, leveraging the same underlying technology that powers consumer chatbots.

A key differentiator for Claude is its "skills" feature, a programmable layer that stores instructions, formatting rules, and even code snippets. By iteratively refining these skills, Shapiro created an automated pipeline that drafts, revises, and formats complex contracts in minutes—a task that previously demanded several hours of senior associate work. Coupled with exceptionally detailed prompts—often 2,000 words long—this workflow produces outputs comparable to mid‑level counsel while freeing lawyers to focus on higher‑value advisory tasks. The virtuous cycle of prompt engineering, skill refinement, and rapid output exemplifies how AI can become a multiplier for firms that invest in process design.

However, the AI revolution is far from uniform. Litigation, criminal defense, and research‑intensive practices confront data‑privacy constraints, model hallucinations, and the need for nuanced argumentation that current LLMs struggle to replicate reliably. Uploading confidential case files to cloud‑based models raises ethical and regulatory red flags, while the risk of inaccurate citations can jeopardize court filings. As a result, firms must adopt a calibrated strategy—leveraging AI for document‑heavy, transactional work while maintaining human oversight in high‑stakes, interpretive tasks. Understanding where AI adds genuine value versus where it introduces risk will determine which law firms thrive in the next decade.