According to the AgentForce Data Configuration and Retrieval Guide, when an organization like Universal Containers wants to enable its AI agent to answer questions using historical case data, the correct implementation is to create an Unstructured Data Model Object (UDMO) based on the Case object, then index that data for retrieval.
The documentation clearly explains:
“When using previous case records to power AI-driven Q & A or similarity-based retrieval, create a UDMO mapped to the Case object. UDMOs allow the system to process and semantically index unstructured text fields such as Case Description, Resolution, and Comments, enabling the LLM to surface contextually similar resolved cases.”
This allows the AgentForce retrieval engine to perform semantic searches across historical support data, returning cases that are most contextually relevant to the user’s query.
Option A is incorrect because past cases cannot be used automatically without indexing them.
Option B is incorrect because a DMO is for structured data (tables, numeric fields) and doesn’t support semantic text retrieval.
Therefore, Option C is correct and aligns fully with Salesforce’s documented best practices.
References (AgentForce Documents / Study Guide):
AgentForce Data Configuration Guide: “Using UDMOs for Case-Based Reasoning”
AgentForce Implementation Handbook: “Indexing Historical Case Records for Semantic Search”
AgentForce Study Guide: “Creating Unstructured Data Model Objects from Case Objects”
