Case Study: Workato achieves 67% lower inference costs and 2-3x faster time-to-value with DigitalOcean NVIDIA H200 clusters

Workato cuts inference costs 67% with DigitalOcean

Workato, a leader in AI-driven enterprise automation, faced significant operational overhead and high costs in managing GPU infrastructure for its AI Research Lab. The lean team needed a high-performance infrastructure partner that could provide rapid deployment and deep technical support to advance their research on autonomous agents, avoiding the slow and generic support of hyperscale providers. They partnered with DigitalOcean to utilize their NVIDIA H200 GPU Droplets and DigitalOcean Managed Kubernetes.

DigitalOcean provided an inference-optimized architecture with high-performance NVIDIA H200 clusters and hands-on engineering support. This solution enabled Workato to achieve a 67% reduction in inference costs, lowering the cost to $0.77 per million tokens. They also accelerated their time-to-value by 2-3x, achieving a 67% throughput increase to 13,561 tokens/sec per GPU and reliable low latency for real-time interactions.

Workato