Interactive Case Study Builder
Case studies that answer back.
A software product for business school faculty. Professors build a company, a cast of AI executive personas, and an assignment, then publish a link. Students consult the executives in conversation and export a report showing exactly how they got to their recommendation.
The problem it solves
The written case study has been the backbone of business education for a century, and generative AI broke it in about eighteen months. A student can paste a case into a chatbot and get a competent answer in seconds. Faculty are left grading work that proves nothing about the student who submitted it.
Banning the technology is not a strategy, and it is not honest preparation for the workplace these students are about to enter. The alternative is to build an assignment where AI is the material rather than the shortcut. If the only way to the answer is to ask good questions of people who will not simply hand it over, then the quality of the questions becomes the thing worth assessing.
What it does
Everything a professor needs to take a teaching goal from a blank page to a published assignment, and everything they need to see what happened afterward.
Design-a-Case wizard
A short guided interview turns a teaching goal into a full draft case: the company, the decision on the table, the cast of executives, and the assignment. Faculty can anchor it to a real-world article or set of notes they paste in, then edit every word of the result.
AI executive personas
Each executive carries a role, a point of view, and information the others do not have. A company holds facts that are freely available and facts that surface only when a student asks a question that genuinely bears on them. Students have to interrogate the room, not prompt it.
Socratic guardrails
Every persona answers substantively and refuses to do the work. No frameworks handed over, no formulas, no step-by-step plans, no checklists. When a student dodges a question, the executive pushes back. The constraint is enforced in the conversation and again in the report, so it cannot be prompted away.
Teaching notes
Faculty get a private answer key generated from the case design itself: the tensions, the defensible positions, the evidence a strong student should surface, and discussion prompts. It is written for the professor and never shown to students.
Evidence gates and PDF export
Faculty set a minimum number of exchanges per executive and a minimum number of executives consulted before a student can export. The export is a PDF containing an analysis of question quality, a summary per executive, and the full transcript.
Faculty keep the judgment
The tool assesses and analyzes for the professor. It never grades. What it produces is evidence of how a student reasoned, handed to the person qualified to weigh it.
How a case runs
- 1
The professor builds the case
Company, executives, and assignment, either from scratch or drafted by the wizard from a guided interview. Faculty decide what each executive knows, what they will volunteer, and what a student has to earn.
- 2
The case is published as a link
One link for the class. Students need no account and no software. The link can be revoked at any time, and the professor keeps a private answer key the class never sees.
- 3
Students interview the executives
Each student works the room in conversation. The executives answer in character, withhold what has not been asked for properly, and push back when a student avoids the hard question.
- 4
The work becomes a report
Once the required depth of consultation is met, the student exports a PDF: an analysis of the questions they asked, a summary of each conversation, and the full transcript. The professor reads the reasoning, not just the conclusion.
Built for an institution, not a demo
Higher education has rules, and a tool that ignores them is not usable no matter how good the teaching idea is. Students sign in to nothing and hand over nothing but the name they type and the questions they ask. Conversation transcripts are purged on a nightly schedule. Each institution connects its own AI provider account, so the model usage sits on its own terms and its own billing rather than on a vendor pool it cannot inspect.
Departments are separate tenants with their own administrators, seats, and settings, and every authenticated request is checked against the department it claims to belong to.
- Platform
- Next.js App Router, React, TypeScript, Tailwind
- Data
- Postgres with a typed ORM, nightly transcript purge on a schedule
- Accounts
- Organization-based authentication, seats, invitations, and roles
- Billing
- Subscription and invoice billing with per-institution seat caps
- AI
- Each institution supplies its own provider key, encrypted at rest
- Quality
- Automated test suite and a typecheck, lint, test, build gate on every change
Why this is on a consulting website
Because it is the honest answer to a fair question. When I tell a client that AI can be built into the way their organization actually works, this is the proof that I do the work myself rather than describing it. The Interactive Case Study Builder is a production product with real tenants, real billing, real legal review, and a real quality gate on every change.
It also shows the judgment I bring to a custom build. The hardest problems in this product were not technical. Deciding what an AI executive must refuse to do, deciding that the tool analyzes and never grades, and deciding what a student should never be asked to hand over are the decisions that made it usable in a classroom. Those are the same decisions I work through with clients building AI into their own operations.
The product is built and in pre-release, with the first classroom pilots scheduled for the 2026 autumn term.
Want Something Like This for Your Organization?
Custom AI tools are built on your infrastructure and owned by you from day one. I teach, you keep.