The AI Proof-of-Concept Challenge

The Mission: From Idea to Investment

Welcome to the central experience of this course: the AI Proof-of-Concept (PoC) Challenge.

For the rest of this semester, your group will operate as a cross-functional startup team within a larger organization. Your mission is to identify a significant business challenge, design an innovative solution using Generative AI, and build a working prototype to prove its value.

The ultimate goal is to create a compelling "launch kit" for your PoC and reflect on your journey in a final team interview. This project is a realistic simulation of how AI products are conceived, built, championed, and reflected upon in a professional environment.

The Process: A Four-Stage Journey

Your team will progress through a structured, four-stage development lifecycle that mirrors modern agile practices. Each stage has specific tasks, goals, and deliverables designed to keep you on track.

1. Stage 1: Ideation & Planning: Define a valuable problem and create a clear, actionable plan and governance model.
2. Stage 2: MVP Development: Build the core, functional version of your chatbot—your Minimum Viable Product.
3. Stage 3: Experimentation & Refactoring: Systematically test, evaluate, and improve your MVP through experimentation and code refactoring.
4. Stage 4: The Launch & Reflection: Create a professional "launch kit" for your PoC and reflect on your learning journey in a team interview.

A detailed week-by-week roadmap with specific tasks and supporting resources is available in the Course Schedule.

Evaluation: The Project Rubric

Your project is worth 45 points of your final grade. It will be evaluated holistically based on the quality of your final prototype, your strategic thinking, and the professionalism of your launch materials. The following rubric will be used for the formal evaluation.

Criteria	Exemplary (A-Level)	Proficient (B-Level)	Developing (C-Level)	Unsatisfactory (D/F-Level)
Business Value & Problem Framing (10 Points)	9-10 Points Project addresses a well-defined, significant business problem with a compelling value proposition. The solution is thoughtfully tailored to a clearly identified user. The project includes a thoughtful and realistic roadmap for future improvements.	8 Points Project addresses a relevant business problem and the value proposition is clear. The solution is appropriate for the target user, though the connection could be stronger.	6-7 Points A business problem is identified, but its significance is unclear, or the AI solution is a weak fit. The target user is defined but the solution is generic.	0-5 Points Project lacks a clear business purpose or solves a trivial problem. The value proposition is missing or confusing.
Technical Execution & Refinement (15 Points)	14-15 Points Achieves all "Proficient" criteria AND successfully implements an optional advanced feature (e.g., RAG, Agentic tools) in a well-integrated and effective manner.	12-13 Points Application MVP is functional, robust, and well-refactored. Code is clean and organized. The team's EXPERIMENT_LOG.md demonstrates a thoughtful process of evaluation and improvement.	8-11 Points Application's core functionality works but may be unreliable. The code is disorganized, and the experimentation process lacks depth or clear evaluation.	0-7 Points Application is non-functional or fails to meet core requirements. The code is difficult to understand or run.
User Experience (UX) & Design (10 Points)	9-10 Points The user interface is intuitive, polished, and professional. The user journey is logical and seamless. The chatbot's persona and tone are consistent and well-suited to the application's purpose.	8 Points The user interface is clean and functional. The application is easy to use with minimal instruction. The chatbot's persona is defined but may have minor inconsistencies.	6-7 Points The user interface is functional but may be confusing or unappealing. The user may struggle with some tasks. The application lacks a clear design or persona.	0-5 Points The user interface is difficult to navigate or non-functional. The user experience is frustrating. No thought was given to design or persona.
PoC Launch Kit & Reflection (10 Points)	9-10 Points The video pitch is highly persuasive and professional. The quick-start guide is exceptionally clear and user-friendly. The team demonstrates deep, critical self-reflection during the final interview.	8 Points The video pitch is effective and clear. The quick-start guide is functional and easy to follow. The team shows thoughtful reflection during the interview.	6-7 Points The video is unclear or unpersuasive. The guide is confusing. The team's reflection in the interview is superficial.	0-5 Points The launch kit is unprofessional, incomplete, or fails to communicate the project's purpose. The team is unprepared for the interview.

Why This Rubric? Connecting Your Work to Real-World Success

The project rubric is more than a grading tool; it’s an "explicit nudge" designed to guide you toward building a product that could succeed outside the classroom. The success of a real-world AI project relies on far more than its technical correctness. It requires a strategic vision, a deep understanding of its users, and a clear plan for adoption and growth.

Here’s how the rubric pushes you to develop these critical skills:

• It forces you to think like a strategist. The Business Value criterion demands that you address a hard problem for the right user group. By requiring a roadmap, it pushes you to think about the project's continual evolution.
• It makes you accountable for adoption. The User Experience criterion directly measures how well you've designed a solution that minimizes disruption for the end-user. An intuitive and polished product is one that people will actually want to use.
• It prepares you to be an organizational champion. The PoC Launch Kit requires you to communicate your project's value not as an academic exercise, but as a real product ready for users. The Team Interview challenges you to articulate your learnings, a key skill for any leader.

By focusing on these four areas, the rubric ensures you are not just learning to code an AI, but learning to build an AI product.