Grounded AI
Grounded AI is an AI system constrained to generate answers only from a defined, retrieved set of source material rather than from its general training data.
Why it matters
Grounding keeps automated security answers tied to your real evidence, lowering the chance of fabricated or off-policy responses that a reviewer would have to catch.
Where it appears in the workflow
Used in questionnaire automation where a model drafts answers from an indexed library of policies, reports, and past responses.
Common mistakes
- Treating grounded ai as a universal term with no workflow context.