Why Law Schools Need This Now
The pressure is real. According to the 2026 AI in Professional Services Report, 77% of professionals expect agentic AI to be central to their workflow by 2030. Law students entering practice in the next few years will be expected to work alongside AI from day one.
The question isn’t whether to integrate AI into legal education. It’s whether you integrate it in a way that builds competence or quietly erodes it.
Prof. Phillips’s approach answers that question directly: design tools that force students to think harder, not less.
The Core Principle: Start With Pedagogy, Not Technology
Here’s the mistake most faculty make. They see a shiny new platform, get excited about the possibilities, and start building before they know what they’re building toward.
Prof. Phillips is blunt about this:
“It can’t just be ‘Let’s use AI!’ There has to be a specific learning outcome.”
Every tool she has built — a case brief helper, a mediation bot, a Bluebook citation assistant, a Socratic method coach, and an employment law counseling simulator — started with a concrete student frustration or a visible gap in learning. Technology came last. The problem came first.
That inversion is the entire framework.
Step 1: Identify a Specific Student Problem
Don’t start with what AI can do. Start with what your students can’t do yet — or what they’re afraid to do.
Prof. Phillips noticed two things: students wanted more practice with Bluebook citations and real feedback, and student anxiety about the Socratic method was actively interfering with their ability to demonstrate legal knowledge they already had. Those observations became tools.
Your starting point should be a specific, observable gap. Not “students need more practice” — but “students freeze when called on in class because they’ve never been wrong in a low-stakes environment.”
Step 2: Design the Interaction Before You Touch Any Platform
Once the problem is clear, map the student experience on paper — or a whiteboard, or a document. Don’t open the tool yet.
Ask yourself: What should the student be able to do after using this? Should they explain a concept, revise a draft, respond to follow-up questions under pressure, or walk through a client intake scenario?
Prof. Phillips describes her process as “thinking about what I want the students to get out of it and then working backwards from there.” This design-first discipline also forces you to define what the tool should never do — which leads directly to the next step.
Step 3: Engineer the Constraints In
This is the most critical step, and it’s where most well-intentioned AI tools fail.
If your tool can give students the answer, it will. And students, under pressure, will take it. That’s not a character flaw — it’s human nature. Your job as the designer is to make that impossible.
Prof. Phillips’s Bluebook Citation Bot will never produce a complete citation on demand. It guides students toward understanding why a citation is structured the way it is. Her Socratic Method assistant is built so the student must drive the conversation — the tool follows, challenges, and pushes back, but it never leads.
The constraint isn’t a limitation of the tool. It is the tool. Build it that way from the start.
Step 4: Try to Break It Before Your Students Do
Before any tool reaches a student, stress-test it aggressively. Feed it incorrect law and see if it pushes back. Probe every path that might accidentally surface an answer. Try to confuse it, mislead it, and exploit it.
“I do a lot of testing and breaking and then rebuilding,” Prof. Phillips explains.
This phase isn’t optional. A tool that fails in front of students doesn’t just create a bad experience — it undermines trust in the entire learning design. Break it privately so it holds up publicly.
Step 5: Pilot Transparently and Iterate With Student Input
When you deploy the tool, tell students exactly what it’s designed to do, what you hope they’ll get from it, and that it may make mistakes. Then invite them to bring those mistakes to you.
This does two things simultaneously. It improves the tool through real-world feedback that no solo testing can replicate. And it repositions students as collaborators in the learning design rather than passive recipients of it.
That shift in dynamic matters more than it might seem.
The Tool That Changed the Most
Of everything Prof. Phillips has built, she considers the Socratic Method assistant the most consequential — particularly for first-generation law students.
The Socratic classroom can feel less like a learning environment and more like a gatekeeping mechanism. Confidence gets mistaken for competence. Students who know the material stay silent because the stakes feel too high to be wrong in public.
“The opportunity to practice being wrong privately is really important,”
Prof. Phillips says.
Students who use the tool arrive in class more willing to participate. Students who use her experiential simulation tools can point to those experiences in job interviews — concrete evidence of skills practiced, not just concepts studied.
That’s the outcome worth designing toward.
What Platforms Do You Actually Need?
Prof. Phillips uses platforms accessible to anyone willing to think carefully about pedagogy. She doesn’t name a single proprietary system as the solution — because the platform isn’t the point.
Tools like custom GPT builders, no-code chatbot platforms, and structured prompt environments are all viable depending on your constraints. What matters is whether the platform allows you to define the tool’s behavior precisely enough to enforce the learning design you’ve mapped out.
If the platform lets AI answer freely without guardrails, it’s the wrong platform for this purpose — regardless of how polished the interface looks.
The Real Barrier Isn’t Technical
The biggest obstacle faculty face is the belief that building AI tools requires technical expertise. It doesn’t.
The hard part — the part that actually determines whether a tool works — is the design. Defining the problem clearly. Mapping the interaction deliberately. Engineering the constraints rigorously. Testing honestly. Iterating openly.
Those are pedagogical skills. Law professors already have them.
Where to Start Tomorrow
If you want to build your first AI practice tool, here’s the shortest possible version of the framework:
- Identify one specific student frustration or skill gap in your current course.
- Write down what you want students to be able to do after using the tool — in one sentence.
- Define what the tool must never do — specifically, what answers it cannot give.
- Map the interaction as a simple conversation flow before touching any platform.
- Choose a no-code tool that lets you enforce those constraints through system prompts or behavior settings.
- Break it yourself before students see it.
- Tell students what it’s for and invite them to report errors.
That’s it. The technology is the easy part. The thinking is the work — and that’s exactly where law professors are already trained to excel.
Comments (0) No comments yet
Want to join this discussion? Login or Register.
No comments yet. Be the first to share your thoughts!