Glossary
What is grounding?
Tying an AI answer to specific source material you supplied.
An ungrounded answer comes from the model's training data, which has a cutoff date and no knowledge of your private pages.
A grounded answer is constrained to documents supplied in the conversation, which makes it checkable: you can look at source two and see whether the claim is there.
Grounding reduces but does not eliminate fabrication. Asking for a source label on each claim makes the remaining errors easy to spot.
In practice
Asking 'using only the pages I sent, which plan includes SSO?' produces a checkable answer; asking it cold does not.
Common questions
- What does grounding mean in one sentence?
- Tying an AI answer to specific source material you supplied.
- Why does grounding matter when you use Claude or ChatGPT?
- Grounding reduces but does not eliminate fabrication. Asking for a source label on each claim makes the remaining errors easy to spot.
- Can you give an example of grounding?
- Asking 'using only the pages I sent, which plan includes SSO?' produces a checkable answer; asking it cold does not.
Related terms
- HallucinationA confident AI answer that is not supported by any real source.
- Retrieval-augmented generationFetching relevant documents and adding them to the prompt before the model answers.
- Context engineeringChoosing and shaping what an AI model sees, rather than rewriting the question.
- Glossary indexEvery AI context term, defined in plain language.
Get LocalBridge
One click sends your open tabs into Claude or ChatGPT. One-time licence, no subscription.
See how it works