Loading
Agentforce and Einstein Generative AI
Table of Contents
Filter by (0)Add
Select Filters

          No results
          No results
          Here are some search tips

          Check the spelling of your keywords.
          Use more general search terms.
          Select fewer filters to broaden your search.

            Search all of Salesforce Help
            Search all of Salesforce Help
            Einstein Data Prism

            You are here:

            1. Salesforce Help
            2. Docs
            3. Quickstart Your Einstein Generative AI Solution

            Einstein Data Prism

            Einstein Data Prism is a grounding solution for generative AI applications within Salesforce, improving the accuracy of AI solutions that use its grounding capabilities. With Einstein Data Prism, automatically ground your large language models (LLMs) so you can gain more accurate and relevant responses to utterances. Einstein Data Prism is automatically enabled in integrated apps, such as Einstein Segment Creation and Agentforce.

            Required Editions

            Note
            Note Prism and Metadata Studio aren’t supported in Government Cloud Plus. Don’t turn on the feature in Government Cloud Plus orgs. Contact your Salesforce account executive for more details.
            Available in: Lightning Experience
            Available in: Enterprise, Performance, Unlimited, and Developer Editions with Foundations, or Agentforce 1 or Einstein 1 Editions

            As part of the Data Prism offering, you can also use the Metadata Studio app. Metadata Studio keeps the human in the loop. It enables the user to improve the results generated by generative AI by validating and contributing business knowledge to the AI-generated metadata descriptions.

            Smart Grounding of Structured Data

            Smart Grounding provides the large language models (LLMs) with the business context to create relevant responses based on accurate and up-to-date metadata, ensuring that outputs are finely tuned to your users’ intents and contexts. Data Prism pinpoints the exact data entities relevant to the natural language question, increasing the accuracy of results and providing greater coverage for large organizations. With Data Prism, ‌generative AI apps, such as Einstein Segment Creation, receive a grounded dataset from the original schema to ensure that accurate data is returned.

            How Does Einstein Data Prism Work?

            After Data Prism is provisioned in your organization, all metadata is scanned (schema, relationships, sample values, and descriptions). All missing descriptions are generated automatically and saved in a vector database. During runtime, when a user asks a natural language question in a Salesforce app, the generative AI app calls the Data Prism API and sends the requested natural language utterance. Data Prism then sends the tables and fields that match the natural language utterance back to the app as focused grounding data. The apps can therefore use this focused grounding data to get more accurate results from the large language models (LLMs).

            Note
            Note The provisioning process can take up to 24 hours to complete.

            To improve the Data Prism results, you can enhance your metadata by enriching the LLM-generated descriptions using the Metadata Studio. Generated descriptions are used in any case, but it’s highly recommended to validate and enrich them to make sure the results are accurate.

            Data Prism Use Cases

            Einstein Data Prism improves the responses and results to utterances or prompts across various Salesforce apps. Here are some examples of how you can use it.

            Segment Creation Based on User Description
            Segment Creation Based on User Description A marketing manager wants to create highly targeted customer segments for a new product launch. They can describe a required target audience in natural language, such as “Create a segment of doctors who are fitness enthusiasts and speak Spanish”. Without Data Prism, it can be hard for the system to identify the correct attributes. With Data Prism, the system identifies and categorizes customers accordingly based on the company’s available data. This enhancement enables marketing teams to tailor their campaigns to unique customer groups quickly without creating overhead on data experts, thus shortening the planning cycle and increasing efficiency.
            • Billing Considerations for Data Prism
              Data Prism uses Data 360 to query sample data and Einstein Requests to call generative AI. Data Prism is called by the Agentforce QueryRecords action and the Data 360 Einstein Segment creation feature. Use of Data Prism impacts the consumption of credits used for billing in these usage types.
            • Metadata Studio
              Use the Metadata Studio to enhance your generative AI results across Salesforce apps that are integrated with Data Prism, such as Einstein Segment Creation. With the Metadata Studio, review and validate all your objects and fields, add missing descriptions, and enhance AI-generated descriptions to gain an accurate semantic representation when querying natural language prompts. If needed, you can also exclude irrelevant objects in the data source so that they aren’t sent to the Large Language Models (LLMs) or returned when grounding.
            • Navigate the Metadata Studio
              Explore the different sections of the Metadata Studio to better understand how to get the best results when validating your metadata.
            • Validate Metadata Entities in the Metadata Studio
              To improve your Einstein results in Data Prism integrated apps, validate an object or field label, by adding or editing an entity description. Standard objects and fields with a description are automatically imported from Salesforce documentation and they’re marked as validated. You can still edit them if you have additional information that’s relevant.
            • Exclude Entities in the Metadata Studio
              Use the exclude function when you want to exclude fields or objects that are no longer in use or are no longer relevant to agent use cases. If fields or objects are excluded, they're not used by Data Prism for grounding purposes. Exclusion only affects solutions that use Data Prism’s API.
             
            Loading
            Salesforce Help | Article