Adopt
Unified API gateway for multiple LLM providers (OpenAI, Anthropic, Google, etc.). Provides a single interface, load balancing, observability, cost monitoring and fallback handling. It is meant to be used for all LLM queries both from the Productboard app and for local experiments.
Exceptions for generating keys directly with LLM providers can be made, but they need a strong rationale (i.e. missing model, specific API, key for a 3rd party service). For local testing purposes we use the staging LiteLLM instance, accessible behind a Tailscale VPN.