How a Law Professor Uses Custom GPTs to Transform Patent Examiner Negotiation Training
Ann McCrackin on bringing realism, feedback, and AI-powered roleplay to the law school classroom
Ann McCrackin teaches skills-based patent practice courses at the University of New Hampshire Franklin Pierce School of Law. In this interview, McCrackin discusses how AI is being used in Patent Practice II.
Source note: This is an edited interview adapted from a narrated video submitted to OpenAI. Watch the associated video in OpenAI Academy.
Intro
Patent law is a field where negotiation, quick thinking, and technical precision meet. For years, Ann McCrackin, a veteran patent attorney and professor at the University of New Hampshire Franklin Pierce School of Law, has sought ways to prepare her students for the high-stakes, unscripted conversations they’ll face as attorneys. Traditional classroom role-plays, she found, fell short—too artificial, too limited, and lacking the expert feedback students need to truly improve. This year, McCrackin turned to AI, building a custom GPT-powered examiner to deliver a leap in realism and learning. In this interview, she shares how she designed the exercise, what changed for her students, and why this approach could reshape skills training across legal education and beyond.
The Interview
Q: Ann, let’s start with the core challenge. Why is negotiating with patent examiners such a critical skill for your students, and what was missing from traditional ways of teaching it?
McCrackin: Negotiating with examiners is at the heart of what patent attorneys do. It’s not just about knowing the law—it’s about persuading someone who’s skeptical, who knows the technical details, and who holds the keys to your client’s patent. Traditionally, I taught this using student role-plays: half the class would play attorneys, half would play examiners. But it always felt artificial. Students playing examiners didn’t really know how examiners behave, so the negotiations weren’t realistic. Plus, only half the students got to practice being the attorney, and I couldn’t give everyone detailed feedback. It was a compromise, and I always wished I could do better.
Q: What inspired you to try using a custom GPT, and how did you go about building it for your course?
McCrackin: I’d been following developments in AI and saw the potential for something much more dynamic. This year, I realized I could use ChatGPT to simulate an examiner—one that would respond in real time, ask tough questions, and give students a true sense of what these conversations are like. I started by designing a detailed persona for the examiner, Carrie McNally, defining her expertise, working style, and authority. I used ChatGPT to help flesh out her character and then built a custom GPT with her persona and all the case documents students would need. The instructions I gave the GPT were intentionally open-ended, so it would react to students’ arguments and make the experience unscripted and authentic.
Q: Can you walk us through what the student experience looks like now, compared to before?
McCrackin: Now, every student gets to play the attorney—they interact with the custom GPT examiner using advanced voice mode, so it’s a real verbal negotiation. There’s no script; the GPT follows their lead, challenges them, and keeps the conversation dynamic. When the negotiation ends, the GPT switches to feedback mode, using a rubric I provided to give immediate, personalized feedback on their performance: how clearly they explained the invention, how well they used claim language, how effectively they distinguished over prior art, and so on. Students can send me a link to their conversation transcript, so I can see exactly how they did. It’s a huge step up in realism, practice, and feedback.
Q: What surprised you most about the impact on your students, and what do you see as the broader implications for legal education or other fields?
McCrackin: The difference was immediate. Every student got a chance to be in the “hot seat,” and the feedback was so much richer and more actionable than anything I could provide in a traditional setting. Students told me it felt real—they had to think on their feet, adapt, and really know their material. From my perspective, I could see into every negotiation, which meant I could spot patterns and coach more effectively. But the broader lesson is that this approach isn’t limited to patent law. Any course where students need to practice interviewing, counseling, or negotiation—anywhere they need to interact and adapt—can benefit from this kind of AI-powered roleplay. It’s a game-changer for skills-based education.
What Stands Out
Core idea: Ann McCrackin’s approach centers on using a custom GPT to simulate realistic, unscripted negotiations, giving every student a chance to practice and receive expert-level feedback.
Classroom design: The exercise replaces artificial student role-plays with a dynamic AI examiner, built on a detailed persona and case files, and uses advanced voice mode for authentic verbal interaction.
Student impact: Students now experience the pressure and nuance of real patent negotiations, with immediate, personalized feedback that accelerates their learning and confidence.
Transferable lesson: This model shows how AI-powered roleplay can transform any skills-based training, offering scalable, realistic practice and feedback in law, business, counseling, and beyond.
Bio
Ann McCrackin teaches skills-based patent practice courses at the University of New Hampshire Franklin Pierce School of Law. She also founded AI-Enabled Attorney to help law firms and law schools put AI to work in ways that make a real difference. With over 25 years of experience as a patent attorney, she brings a mix of deep legal expertise and innovative thinking to both the classroom and the legal profession.