Launched LLM-as-a-judge evaluations on real production traffic. Create evaluators using built-in templates or custom judge prompts, and Grepture scores actual responses on a 0-to-1 scale. No synthetic datasets or separate eval pipelines needed.
Read the full post