Legal prompt engineering
Prompt engineering is the art of crafting precise, structured instructions to get optimal results from a generative AI. It is not programming but a methodical way of formulating requests in natural language. The ABA (2025) views prompt engineering as a natural extension of a lawyer's existing skills.
Prompt engineering has become an essential skill for legal professionals using generative AI. The idea is to structure your requests so you get accurate, relevant and actionable answers. Several techniques exist: zero-shot (a direct instruction with no example), few-shot (with examples), and chain-of-thought (step-by-step reasoning), the latter being especially well suited to legal reasoning.
A good legal prompt combines four key elements: the role ("You are a lawyer specialising in employment law"), the context (jurisdiction, type of client, what is at stake), the task ("Draft a summary note on...") and the expected format (structure, length, level of detail). This methodology turns a generic tool into a high-performing legal assistant.
In 2025, the American Bar Association recognised that prompt engineering amounts to a natural extension of a lawyer's skills. Asking the right questions, structuring an argument and assessing whether an answer holds up are reflexes already deeply embedded in legal practice. Zevra's training programmes teach these techniques adapted to the French legal context.
Related terms
An LLM (Large Language Model) is an artificial intelligence model trained on massive volumes of text to learn the statistical stru…
AI hallucinations refer to the generation of false, fabricated or inconsistent information presented with apparent certainty by a …
Temperature is a hyperparameter that controls how much randomness goes into an LLM's responses. A low value (close to 0) produces …
AI-powered legal automation refers to using artificial intelligence technologies to automate repetitive, low-value tasks: reading …
An AI agent is a new generation of autonomous AI system able to perceive its environment, set goals, plan actions and act independ…