RAG looks up relevant documents and writes an answer from them. An AI agent can plan, use tools, and take several steps, such as querying logs, reading code, and checking a customer's config, before it answers. RAG answers what's written down; an agent can investigate what isn't.
When RAG is enough
- How-to and policy questions
- Product questions covered in docs
- Questions where the answer is the same for every customer
When you need an agent
- Errors and bugs specific to one customer
- Issues introduced by a recent deploy
- Anything support would normally escalate to engineering
Decimal uses retrieval for documented answers and investigation for the rest. See RAG and agentic investigation.