The Code env and Container Execution used by Retrieval-Augmented LLM are the same as those used for the Knowledge Bank.
Go to the Knowledge Bank - {{retrievableKnowledge.name}} to edit these properties.

Completion settings

General guardrails

Advanced

Minimum retention time

In seconds. Ensures at least one retrieval-augmented LLM instance is retained for this duration after usage. This helps reduce latency on intermittent requests by avoiding startup delays.