Grounding
Supplying an LLM with authoritative, current, customer-specific data inside the prompt so its response is anchored in real information, not training data.
Quick answer
What is Grounding?
Supplying an LLM with authoritative, current, customer-specific data inside the prompt so its response is anchored in real information, not training data.
Last updated:
Grounding is the antidote to hallucination. Without grounding, an LLM answers from its training data — which may be months or years out of date and contains zero knowledge of your customers, products, or processes. With grounding, you inject the relevant facts into the prompt: account history, latest policy docs, current pricing.
In Salesforce, grounding sources typically include: Account/Contact/Case records, Knowledge Articles, Opportunity history, related documents in Files or Content. The Einstein Trust Layer and platforms like gptfy automate grounding by retrieving relevant records (via RAG), formatting them into a system prompt, and ensuring the LLM only answers from grounded context.
A well-grounded prompt dramatically reduces hallucination but increases token cost (more context = more tokens). The art is in retrieval — pulling exactly the relevant grounding without flooding the prompt.
Related terms
Browse all terms- RAG (Retrieval-Augmented Generation)An LLM is given relevant retrieved documents as context before generating a response — grounding outputs in your specific data, not just the model's training.
- Hallucination (AI)When an LLM produces output that sounds plausible but is factually wrong or fabricated — e.g. citing a non-existent record or inventing a policy detail.
See it in your Salesforce org
See Grounding running in GPTfy
Book 30 minutes with a GPTfy engineer to see how Grounding actually works inside a Salesforce org like yours.
Book a demo