Why Smart Lawyers Are Building AI Tools Instead of Buying Them

The legal profession is moving beyond the simple choice of buying or avoiding AI, with a growing trend of lawyers building their own AI tools through a method called “vibe coding.” This approach, which involves directing AI systems with natural language feedback, lowers the technical barrier to AI development and allows for rapid experimentation. The article argues that for many tasks, current large language models (LLMs) already perform at or above the level of human lawyers, making custom-built tools a viable alternative to expensive, rigid vendor solutions.

These custom tools, created in platforms like ChatGPT or Gemini, offer advantages like encoding specific lawyer judgment, rapid customization, and a low-cost environment for testing workflows and training. They are particularly useful for simulations, such as mock arguments and witness examinations, and for scaling a firm’s expertise by capturing the methods of its most effective lawyers. Building tools also fosters 'ethical competence' by forcing lawyers to understand how AI systems function and address potential risks like hallucinations.

While vendor solutions remain appropriate for comprehensive practice management and enterprise systems, the article advocates for a balanced approach. Law firms should explore “what can we build?” alongside “what should we buy?” to leverage the benefits of both approaches and position themselves for a future where AI literacy is essential for legal practice.

(Source：National Law Review)