“Is Data Cloud Required for Agentforce?” is a question I often come across on various social media platforms. This blog post is my attempt to demystify this,
offering insights to help you gain clarity and make informed decisions.
The Simple Answer
No, Data Cloud is not required to use Agentforce. You can leverage Agentforce’s core functionalities without having Data Cloud.
Update (04-Feb-2025): This generated some controversy on my LinkedIn post, with several individuals pointing out that Data Cloud is mandatory for using Agentforce. To clarify, I contacted Salesforce Support (Case #469613511), and the Data Cloud Technical Support Engineer confirmed that Data Cloud is not mandatory for Agentforce. I have included a screenshot from the support case below for reference.
Nonetheless, please double-check with your Salesforce Account Executive on this matter.
But…
While not mandatory, Data Cloud complements Agentforce by enhancing its features and functionalities in several ways:
1. Provide More Context & Get Better Responses
Data Cloud enables seamless integration of data from external systems into Salesforce. This additional context can enrich Agentforce’s capabilities. For instance, you can ground your prompt templates using data model objects (DMOs) from Data Cloud in Prompt Builder, leading to more relevant and precise AI responses.
2. Retrieval Augmented Generation (RAG)
RAG boosts the quality, accuracy, and relevance of responses generated by large language models (LLMs). Using retrievers with Einstein Search, you can enhance your prompt templates by integrating knowledge from unstructured sources such as articles, emails, and chat transcripts.
The Resource field in Prompt Builder contains active retrievers that you can add to prompt templates. The default retrievers are created when a search index is created in Data Cloud.
Reference help article: Ground with Retrieval Augmented Generation (RAG) in Data Cloud
3. Einstein Data Library
Assigning data libraries to Agents improves accuracy, enhances personalization, and builds trust. The Einstein Data Library indexes knowledge articles, fields, or file uploads, creating a searchable index that ensures LLM responses align with your organization’s information.
When you create a data library it automatically pushes data streams or files to Data Cloud, creating a search index and retriever, and linking the agents to that data.
Refernece help article: Einstein Data Library
4. Einstein Trust Layer – Audit Trail
The Audit Trail tracks generative AI usage in your Salesforce org, ensuring compliance with security, privacy, regulatory, and AI governance policies. The audit data, along with feedback, is stored in Data Cloud.
Reference help article: Audit Trail
5. Einstein Trust Layer – Verify Masked Data
Data Cloud supports monitoring data masking configured in the Einstein Trust Layer. Build Data Cloud reports to verify and monitor that sensitive data is appropriately masked in your LLM prompts.
Reference help article: Verify Masked Data
6. Einstein Trust Layer – Review Toxicity Scores
You can use Data Cloud to build reports that review toxicity scores of LLM-generated responses. These reports help ensure that the AI aligns with organizational standards.
Reference help article: Review Toxicity Scores
7. Einstein Audit and Feedback Data
Collect and store Einstein generative AI audit and feedback data in Data Cloud to refine applications through feedback analytics, prompt adjustments, and LLM tuning. Enabling this feature requires Data Cloud.
Reference help article: Einstein Audit and Feedback Data
8. Agentforce Analytics
Agentforce Analytics uses Data Cloud data to help you gain insights into how your agents are performing. The data in the customizable dashboard and reports can help you determine whether you want to change anything to drive adoption or adjust topics and actions to improve user satisfaction.
Reference help article: Agentfoce Analytics
