A strong prompt is only the starting point. Better AI results depend on giving the tool a clear role, action, context, expectations and restrictions, then improving the response through feedback, iteration and verification. In this practical Sensemaker Academy taster, Factor’s Nicole Bradick and Michelle Sherman show how legal professionals can build those habits into their day-to-day work.
The session explores the RACER framework for clearer prompting and shows why the first answer should rarely be the final one. Michelle explains how targeted feedback, alternative options and clarifying questions can improve an initial response. She also explains why AI can produce fluent but incorrect answers, where hallucination risk is higher and how to manage the “verification tax” without simply redoing the entire task.
Using a guided sandbox exercise, participants practice asking AI to show the source text behind its conclusions, assess how well its claims are supported and pressure-test its answer for errors or alternative interpretations. The workshop closes with an introduction to vibe coding, showing how plain-language instructions can turn an idea into a simple prototype. The purpose is not to turn lawyers into developers, but to help legal professionals explore what they could build or improve within their own workflows.
• How to use the RACER framework and structured iteration to give AI clearer instructions and improve its first response
• When AI is more likely to produce incomplete or inaccurate answers, and three practical techniques for verifying its work efficiently
• What vibe coding is, and how legal professionals can use plain language to prototype and refine simple tools and workflows