Giskard Hub

Giskard Hub is an AI security and continuous testing platform for Conversational AI agents. Operating as a black-box testing tool, it evaluates text-to-text agents via their API endpoint and runs automated red-teaming, functional tests, and vulnerability scanning to detect issues such as prompt injection, sycophancy, data disclosure, inappropriate content, hallucinations, contradictions, omissions, and inappropriate denials. Each assessment yields a full vulnerability report with ranked findings, a go/no-go deployment recommendation signed by AI security experts, and remediation guidance. Giskard also manages remediation: qualifying findings, prioritizing fixes with your team, opening tickets in your workflow, and re-running tests to confirm fixes. The platform supports continuous monitoring and proactive generation of attack scenarios (leveraging internal knowledge like RAG bases and external resources), collaborative red-teaming playgrounds and annotation tools for business users, and can be deployed on-premise for sensitive workloads. Data residency choices (EU or US), end-to-end encryption, RBAC, audit trails, integration with identity providers, and a 0-training policy are emphasized to protect customer data. The company also offers an Open-Source tier and a Hub enterprise tier; a refund for the assessment is available under a limited-time offer if no vulnerabilities are found.