Winning Microsoft's AI Hackathon: A Multi-Agent Marketing Automation Journey

I wanted to share my experience winning third place in Microsoft's AI hackathon—complete with demo and code included in the full article!

What We Built: A multi-agent marketing and campaign automation system designed for small to medium-sized enterprises. Here's how it works:

The system analyzes customer data and generates personalized messages using GPT models. Each agent has its own specialized function and AI model. These messages then pass through Azure's Content Safety Agent, which flags any harmful, hateful, violent, or sexual content. A fifth orchestration agent tracks progress across all agents and manages message delivery to customers.

This multi-agent system not only helps businesses retain loyal customers but also eliminates the need for constant human oversight while ensuring compliance safety is mandatory—not optional.

The Challenge: We had just two weeks from start to finish. The winning criteria required at least 25% of a working project.

The Journey: Honestly? I almost gave up. The team members who initially pushed me to join this hackathon (one I'd never even heard of!) weren't delivering the results I hoped for. But now I feel blessed and deeply thankful that I accepted the challenge, took responsibility as the leader, and submitted our demo—even though it was far from perfect.

That "broken demo" earned us third place.

The Real Prize: Sure, the monetary prize was modest, but the real value? The journey. The learning. The experience of working with complete strangers and ultimately leading them to success. That's priceless.

I've written a detailed article about this experience. I know posting links might not be allowed here (which I completely respect), but I couldn't stop myself from sharing this milestone with the community!

Thank you for reading my story!

6 comments