LLM Judge checks: Evaluate anything you can describe in words | Giskard Hub v.3.0

Some checks are trivial to script: the response is valid JSON, the amount matches, the field exists. Whether the answer stayed grounded in your knowledge base, or refused for the right reason, has no regex. Giskard Hub v3 hands those to an LLM judge you configure yourself. 🎯 What's new: - LLM Judge check — Describe the pass condition in plain language. Craft your own prompt template and evaluate a single interaction or the full trace against it. - A full evaluator library — String matching, comparison, and structured or automated validation, for the assertions that don't need a model to decide. - Custom checks — Compose your own evaluators on top of the existing ones instead of maintaining test logic outside the platform. - Applies anywhere in the trace — Run any check on the final answer or at an intermediate step of a multi-turn conversation. 📖 Documentation: https://docs.giskard.ai/hub/ui 🚀 Request a demo: https://www.giskard.ai/contact #AIsecurity #LLMEvaluation #LLMasaJudge #AIagents