Product tour

Detect and Prevent Adversarial Attacks on LLM Applications

Start the self-guided tour to learn how to:

  • Detect hallucinations causing changes in a LLM chatbot's behavior
  • Discover attempted prompt injection attacks
  • Visualize prompts and responses clusters in a 3D UMAP visualization
  • Perform root cause analysis to identify malicious users and prompts
  • Create a custom metric and alert to immediately identify future attacks