Get Policy Insights with Einstein

Let Einstein use the information generated by your agent actions to create insights into your policies in plain language. Get detailed insights for policy summaries generated with the Get Policy Details Agent Action. Or get one-sentence insights for policy lists generated with the Get Policies by Object and Get Policies by Policy Type Actions.

Required Editions

Available in: Enterprise, Performance, Unlimited, and Developer Editions with the Einstein Platform add-on
User Permissions Needed  
To execute actions in Agentforce (Default): Use Agentforce Default Agent
To run prompt templates: Execute Prompt Templates
To view Policy Center policies: View All Policy Center Policies
To create and edit Policy Center policies: Modify All Policy Center Policies
  1. From Setup, in the Quick Find box, search for and select Prompt Builder. Then click New Prompt Template.
  2. Under Prompt Template Type, select Flex.
    New prompt template with Policy Insights details
  3. Give your template a name.
    • For detailed summaries that use information generated by the Get Policy Details agent action, we suggest the template name Policy Insights.
    • For short summaries that use information generated by the Get Policies by Object or Get Policies by Policy Type agent actions, we suggest the template name Short Policy Insights.
  4. Under Define Sources, enter the name Policy Detail.
  5. For Source Type, choose Free Text. Then click Next to go to the Prompt Template Workspace.
    Prompt Builder workspace with Policy Insights instructions
  6. Add instructions for the LLM that define what to include in the response and conversation guidelines. Our suggested instructions should be updated and modified as you continue to use Einstein to get the best results.
    • For Policy Insights generated from the Get Policy Details agent action, we suggest:
      Take the {!$Input:Policy_Details}, which are the technical details about a data
             security and privacy policy, and rewrite them in simple, easy-to-understand language. The
             explanation should be clear, concise, and accessible to non-technical users. Focus on what
             the policy does, and why it matters, and how it assists the company with security and privacy
             needs. Avoid jargon and use relatable examples when needed. Keep the response engaging,
             reassuring, and straightforward.
    • For Short Policy Insights generated from the Get Policies by Object or Get Policies by Policy Type agent actions, we suggest:
      Take the {!$Input:Policy_Details}, which are the
             technical details about a data security and privacy policy, and rewrite them in simple,
             easy-to-understand language. The explanation should be clear, concise, and accessible to
             non-technical users. Focus only what the policy does. Avoid jargon and use relatable examples
             when needed. Keep the response engaging, reassuring, and straightforward. The response should
             be no longer than one sentence. The name of the policy should be included as the first words
             of the response in bold text.
  7. To test the Policy Insights prompt template, click Test Inputs, then add the policy details previously generated from the Get Policy Details action.
  8. To see how your changes to the prompt template impact the LLM response, click Preview. See Review, Revise, and Repeat: Ensure That a Prompt Template Is Effective.
  9. When you’re satisfied with the outputs, click Activate.
    If there are multiple available versions of a template type, Einstein uses the most recently modified active template version when drafting summaries.