Case Study: Simli achieves real-time, cost-efficient AI avatar inference with DataCrunch

How Simli Achieved Cost-Efficient, Real-Time Inference for Interactive AI Avatars

Simli, a developer of interactive AI avatars, faced the challenge of delivering lifelike, real-time digital experiences that required ultra-low latency and production-grade stability, all while remaining cost-efficient for a startup. They needed a infrastructure partner that could provide this reliability without the high cost of traditional hyperscalers.

DataCrunch provided Simli with bare-metal GPU clusters and on-demand GPU resources, which were specifically configured for their real-time inference workloads. This solution resulted in 30-50% faster GPU startup times and allowed Simli to achieve 2–3 times more avatar sessions per dollar. By utilizing DataCrunch, Simli met its sub-300ms latency requirement and significantly reduced costs, enabling them to scale their interactive AI API service effectively.

Simli

Lars Vagnes

Founder & CEO