Building an AI-Powered MVP in 48 Hours
How our team beat 85 global MBA competitors by focusing on a hyper-specific user pain point and leveraging AI for rapid development.
Key figures
The Challenge
The INSEAD Product Games is one of the world's largest MBA product competitions. The prompt: "Build a product that leverages AI to solve a real user problem." Constraints: 48 hours, working MVP required, judged by product leaders from Google, Amazon, and Grab.
The Problem: "Wait, what did we agree on?"
We interviewed 15 project managers and found a recurring pain point: effective meetings often result in unclear next steps. While transcription tools exist, they generate walls of text that no one reads. The gap wasn't in recording the meeting, but in structuring the output into action.
The Approach
With limited coding experience in the team, we used AI (Claude 3.5 Sonnet & Cursor) as our primary developer — allowing us to focus on product strategy and user testing.
- Hour 0–6: User interviews & problem definition
- Hour 6–24: Prototyping the "Action-First" interface
- Hour 24–36: Building the MVP using Python/Streamlit + OpenAI API
- Hour 36–48: User testing & pitch deck refinement
The Solution: ActionItem.ai
We built a tool that didn't just transcribe, but extracted and assigned. Users could upload audio, and our system would identify:
- Decisions Made (vs. just discussed)
- Owners Assigned (who is doing what)
- Deadlines (implied or explicit)
The Outcome
We won the Grand Prize out of 85 teams. The judges cited our "radical focus on the last mile of meeting productivity" and the functional completeness of our MVP. This project proved that with the right product thinking, technical barriers to entry are lower than ever.