Frugal has two approaches to optimization - manual and automated. For manual optimization, we point engineers at the components, call sites and queries driving the highest spend. Using this data, engineers can manually rearchitect the most impactful code or features.
For automated optimization, Frugal reviews code and cloud configuration for over 180 different types of waste across all your costs including: AI, storage, logs, metrics, traces, serverless, databases and Kubernetes.
We find waste and produces code fixes for issues such as:
Rebalancing API costs to storage costs
Rightsizing high-cost/low-value logs and metrics
Reducing AI token consumption with prompt caching and structured outputs
Applying caching and batching to offload frequent, costly usage
Rightsizing Lambda functions or Kubernetes pods.
We also address best practices—tagging untracked costs, assigning clear code ownership, and ensuring cost visibility.