Agentforceis built to be dynamic, meaning it doesn't just follow a static script but rather "reasons" through a problem using an agentic loop. The feature that directly allows an agent to adapt its behavior and take specific, real-time steps based on customer input isCustom Actions3. When a customer provides information—such as a specific order number or a request to change an address—the agent identifies the user's intent and selects the most relevant action to execute4. These actions can be powered by Flows, Apex, or Prompt Templates, giving the agent the ability to interact with Salesforce data in real-time. While Data Cloud (Option A) provides the data foundation for grounding, it is the actions themselves that define the agent's behavioral capabilities6. Screen flows (Option C) provide a guided UI for humans, but custom actions are what allow the agent to function autonomously7. Tableau (Option D) is an analytics tool and does not drive real-time agent behavior8.