...The system integrates directly with local or remote Ollama servers, enabling seamless access to models already deployed in a user’s environment. It also includes experiment logging and A/B testing capabilities, which allow users to compare outputs side by side and track performance metrics such as latency or token usage.