What it is
Grounding is the technique of connecting an LLM to real data sources — your CRM records, knowledge articles, product catalogs — so its responses are based on facts rather than training data. In Salesforce, grounding connects Agentforce to Data Cloud, Knowledge, and custom objects.
Why it matters
Without grounding, AI agents hallucinate — they make up plausible-sounding answers that are factually wrong. Grounded agents cite your actual data, which means customers get accurate answers and your team can trust the output.
Key components
- Data Cloud integration
- Knowledge article indexing
- Semantic search
- Citation tracking
How it connects
Agentforce uses Data Cloud and Salesforce Knowledge as grounding sources. RAG is the primary grounding technique used under the hood.
Good to know
Grounding quality depends on data quality. If your Knowledge articles are outdated or your CRM data is messy, your grounded agent will confidently repeat bad information.
Related terms
RAG (Retrieval-Augmented Generation)
A technique that makes AI smarter by fetching relevant information from your data before generating a response. The AI "looks it up" instead of guessing.
Knowledge Management
The systems and processes for organizing, storing, and making business information accessible to AI agents so they can provide accurate, contextual answers.
Data Cloud
Salesforce's unified data platform that centralizes customer data from all sources, making it available for agents, analytics, and insights in real time.
Trust Layer
Salesforce's AI trust and safety framework that ensures agents operate within compliance boundaries, prevent misuse, and maintain data security.