Case Study: Toma achieves 75%+ call resolution and lower latency with Porter and Deepgram

TOMA cuts EC2 spend by 50% and reaches 75%+ call resolution with Porter

The customer, TOMA, a startup providing voice AI solutions for automotive dealerships, faced significant challenges with reliability, latency, and control while using shared infrastructure from Railway. These issues culminated in a service outage during a crucial demo, highlighting the unsuitability of their platform for mission-critical, real-time voice processing. This prompted TOMA to seek a new vendor, Porter, for a managed Kubernetes solution.

Porter provided TOMA with managed Kubernetes infrastructure on AWS, offering the ideal balance of control and simplicity. The migration was completed in just two days. By using Porter to self-host Deepgram's Nova-3 speech-to-text model on optimized, colocated infrastructure, TOMA dramatically reduced latency and improved transcription accuracy. This solution resulted in call resolution rates exceeding 75% for their dealership customers and cut their monthly EC2 spend from over $13,000 to approximately $6,000.

TOMA