Giskard

Giskard is an automated testing platform for LLM agents that helps enterprises continuously detect vulnerabilities such as prompt injection, data disclosure, hallucinations, and sycophancy. It operates as a black-box testing tool, requiring only an API endpoint to assess conversational AI agents in text-to-text mode. The platform offers continuous red teaming, LLM evaluation, and AI guardrails, with both an open-source solo tier and an enterprise Hub. Key features include automated vulnerability detection using internal knowledge bases and security taxonomies, proactive monitoring throughout the AI lifecycle, and a collaborative red-teaming playground accessible to business stakeholders. Giskard supports data residency and isolation (EU or US), role-based access control, and on-premise deployment for sensitive workloads. It is trusted by enterprise AI leaders such as L'Oréal, AXA, Société Générale, and Google DeepMind.