# What is Grounding?

> Tying an AI's answer to specific source material it can cite.

Canonical: https://www.decimal.app/glossary/grounding · Last reviewed: 2026-09-23

Grounding means an AI model's output is based on, and traceable to, real sources such as documents, logs, code, or records, rather than on what the model learned during training.

## Why it matters

Support answers have to be right for this customer, this version, this config. Grounding is what makes an AI answer checkable and keeps it from inventing plausible details.

## Best practices

1. Cite inline. Link each claim to its source.
2. Prefer primary sources. Code and logs beat a summary written last year.
3. Say 'I don't know'. An answer with no source behind it shouldn't be sent.

## How Decimal helps

Every Decimal answer and investigation carries citations to the code, logs, tickets, or docs it used.

## Related terms

- [Hallucination](https://www.decimal.app/glossary/hallucination): When an AI model states something confident and wrong.
- [Retrieval-Augmented Generation (RAG)](https://www.decimal.app/glossary/retrieval-augmented-generation): Having an AI model look up relevant documents before generating an answer.
- [AI Agent](https://www.decimal.app/glossary/ai-agent): Software that plans, uses tools, and acts toward a goal instead of answering one prompt.

---

Decimal · Book a demo: https://cal.com/decimal-demo/30m · Markdown index: https://www.decimal.app/llms.txt