Large undergraduate classes create particular teaching challenges. A daunting challenge is providing personalized, meaningful learning support at a scale that meets the needs of hundreds of students. In subjects like law that rely heavily on conceptual understanding and applied analytical thinking, this need for support is even more pronounced. Undergraduate business students often enter a law course with little familiarity with legal terminology and disciplined reasoning.
This past fall, I taught a business law course to all second year students in the Bachelor of Commerce program at Dalhousie University. I had 343 students across three sections of 109 to 120 students in each. The course covers foundational areas of Canadian business law and requires students to apply that law with a structured legal analysis. Even with active learning approaches in class and clear instructional structures, it was apparent that students needed individualized, on-demand support that traditional office hours and T.A. tutorials could not fully satisfy.
To address this, I created and deployed two custom AI tutors, Harvey and Elle, built as custom GPTs in the ChatGPT platform. The aim was to offer scalable, digital learning companions that aligned directly with course learning outcomes and pedagogical needs. What emerged was an effective model for AI-supported instruction that helped students better understand legal concepts, improve their analytical skills, and engage more confidently with course material.
Why Build Custom AI Tutors?
Although generative AI tools like ChatGPT, Gemini and Claude are widely accessible, they present limitations that can make them inappropriate for academic use by students, especially in law courses. Generic large language models often produce:
Full blown “answers” that shortcut students’ learning processes, U.S.-based legal explanations rather than Canadian ones, Oversimplified or incomplete reasoning, and Explanations that do not follow the course’s terminology and required analytical structure
For these reasons, I needed custom AI tools that were aligned with the Canadian legal framework, the course materials, and the expectations for structured analysis. Just as important, the tools needed to explain and teach, not simply provide students with answers to be used for assessments. At the same time, the AI tutors needed to be freely accessible, easy to use, welcoming, and engaging.
Two Tutors, Two Roles
A key design decision was to create two separate AI tutors, each with a clear instructional focus.
Harvey – Your Business Law Tutor was designed to help students understand legal concepts. He provides:
Plain-language explanations of legal rules, Examples to reinforce understanding, Guiding questions that help students reflect on their learning, Clarifications of terminology and principles, and Practical, real-world business implications
Elle – Your Legal Case Analysis Coach supports students in practicing the three-step legal analysis required for course assessments. She guides students through the process of: (1) identifying the legal issue, (2) setting out the applicable law, and (3) applying the law to the facts of a case.
Elle asks a student to input a case fact situation (usually from class and course materials) and, importantly, that student’s own attempt at analyzing the case. She does not rewrite the student’s analysis, but provides a constructive critique that helps the student improve their analysis by identifying missing steps, improving organization, and offering reminders about the importance of making connections between facts and relevant law.
These two roles reflect my course design and pedagogical approach of scaffolding conceptual understanding with analytical application. In other words, students must understand before they can analyze. Harvey and Elle were built to support both of those phases of learning.
To ensure consistency with the course materials, Harvey and Elle were “trained” by uploading most of the course materials which included instructor-created content such as slide decks, the syllabus, and YouTube lecture videos. While uploading the course textbook would have been ideal, it was not possible due to copyright restrictions.
Humanizing the AI Tutors
Each AI tutor was given a distinct, approachable personality. Both Harvey and Elle were inspired by recognizable lawyer characters in pop culture. Harvey is a gritty, shrewd, Bay Street (the Canadian equivalent of Wall Street) lawyer. Elle is witty and bubbly with a sharp legal mind. The goal was to reduce the intimidation students sometimes feel toward legal subjects and create an environment where they felt more comfortable asking questions. Harvey and Elle needed to be relatable and approachable.
The personalities served as learning cues. Harvey’s confident, polished, explanation-centered style signaled that he was the right tutor for clarifying confusing concepts. Elle’s supportive, energetic, coaching-like tone made her suited for guiding students through legal reasoning. These personalities helped humanize the AI tools, making interactions welcoming, engaging and, sometimes, humorous.
Student Use and Perceptions of the AI Tutors
Students were surveyed at the middle of the semester after the midterm exam. Almost a third of all students in the course responded. Their feedback highlighted several patterns in how and why students used the tutors.
1. Regular and Voluntary Use
Even though the use of the AI tutors was optional, almost all survey respondents (95%) had engaged with one or both tutors. One third used at least one tutor a few times per week. The small minority (5%) who did not use any of the tutors felt they were unnecessary for their learning.
2. Clear, Understandable Explanations
The most common theme in students’ comments was appreciation of how Harvey and Elle explained legal concepts in ways that were easy to understand. Comments frequently mentioned:
Clearer explanations than other resources, Help in making legal concepts more digestible, and Improved understanding of terminology
3. Support with Legal Analysis
Many students valued Elle’s ability to guide them through the three-step legal analysis, noting that she helped them structure their responses and understand how to apply legal rules to fact patterns. This was especially important for midterm test preparation.
4. Role of Each Tutor
Survey responses indicate that students used Harvey primarily for understanding legal concepts and Elle primarily for practicing structured legal analysis. This reflects the distinct instructional roles the two tutors were designed to play.
5. Ease of Use and Accessibility
A number of students valued that the tutors were easy to access, available anytime, and convenient for quick clarification.
6. Help Through Examples and Practice
Some students noted that they liked how the tutors provided examples or helped them practice course concepts.
7. Confidence, Motivation, and Engagement
Most students reported feeling ‘very confident’ or ‘somewhat confident’ in the accuracy of Harvey and Elle’s responses. Many also indicated that using the AI tutors increased their motivation and engagement in the course. Many used the tutors when preparing for the midterm, suggesting that having immediate feedback contributed to a greater sense of readiness.
8. Custom AI vs. General AI Tools
Half of respondents found Harvey and Elle more helpful than non-custom AI tools such as ChatGPT, Gemini and Claude. Many students liked that Harvey and Elle were custom designed for the course with content consistent with course materials and learning outcomes.
9. Teaching vs. Answers
Many students appreciated that Harvey and Elle guided them in learning, without simply providing answers. Other, but fewer, students were frustrated that the AI tutors did not provide straightforward answers. Almost all respondents (93%) agreed that the AI tutors helped them to learn more effectively, instead of removing or lessening the need for them to learn.
Conclusion
Integrating Harvey and Elle into a large business law course significantly enhanced students’ ability to understand and apply legal concepts. The tutors offered accessible, personalized learning experiences that supplemented instruction in ways not feasible at scale with traditional methods.
For educators exploring AI in their own teaching, purpose-built AI tutors with clear roles, customized content, pedagogical grounding, and approachable personalities can offer a potentially powerful, scalable model for supporting student learning.
Wayland Chau, BCom, JD, LLM, is an Assistant Professor in the Department of Strategy, Entrepreneurship and International Business of the Faculty of Management at Dalhousie University, where he teaches business law and develops AI-enhanced approaches to teaching and learning. A former practicing lawyer with extensive experience in the financial sector, he now focuses his teaching and consulting on generative AI, active learning, core competency building, law, ethics, and social responsibility. He also shares reflections on teaching and learning through his blog, The Reflective Prof.
