The Challenge
Swing connects schools with substitute teachers—a classic two-sided marketplace. When I joined, the matching process was heavily manual. Operations costs were eating into margins and making the unit economics unsustainable.
What I Built
AI-powered matching - Replaced manual matching with an ML system that considers teacher qualifications, school preferences, location, and historical performance. The system gets smarter with every placement.
Automated workflows - Eliminated manual touchpoints throughout the booking and payment process. What used to require human intervention now happens automatically.
Quality feedback loops - Built systems to capture and act on quality signals, improving match quality over time.
The Results
- 50% OpEx reduction in one year
- Sustainable unit economics achieved
- Faster matching with higher quality
Key Lesson
The best AI implementations aren’t about replacing humans with robots—they’re about encoding human expertise into systems that scale. The matching algorithm captured what our best operators knew intuitively.