Close this search box.
Predictive and GenAI with Data Cloud

Unlocking Predictive and Generative AI with Einstein Studio and Salesforce Data Cloud

As Salesforce doubles down on making AI more accessible to all, Data Cloud customers can now take advantage of predictive analytics and generative AI functionality like we’ve never seen before with Einstein Studio.

Einstein Studio provides an expanding palette of tools for implementing predictive AI/Machine Learning models at varying levels of complexity. It’s also the control center for Generative AI models, to enable teams with AI-powered insights across their workflows.

Putting predictive analytics to work

For Data Cloud users who want to utilize predictive analytics, there are various tools you can leverage to build, deploy, and manage AI/ML models. Einstein Discovery is now available within Einstein Studio as the No-Code Model Builder, which can be used to train and deploy models using Data Cloud data.

If you want to take a more custom approach, Einstein Studio also allows for models to be built in external cloud-based platforms such as Databricks, Amazon Sagemaker, and Google Vertex. If you’re already using those tools, this functionality allows you to more easily integrate your existing models with Data Cloud — and, by extension, Salesforce.

Training models is just one part of the equation. There are various ways that organizations can make predictions on Data Cloud data. The first option, using Batch Data transforms to score many records at a time, is valuable for quick evaluation of models during the buildout and validation process, and can be useful if batch scoring is necessary.

For more automated ways to consume predictions, you can activate a model and use Einstein Studio prediction jobs and Flow Builder actions to surface model outputs directly in Salesforce. Models don’t provide value unless the insights and predictions they generate are operationalized by end users to make better decisions, so this functionality is a key driver of value.

What’s new with generative AI?

Over the past year, Salesforce has been working to drive new functionality via the use of the Einstein 1 platform; specifically, Einstein Copilot and Prompt Studio.

Prompt Builder provides organizations with a framework and tools to create, test, and operationalize prompts to Generative AI models like GPT-4. Einstein Copilot provides a conversational interface embedded into any app in Salesforce that can carry out “actions” like summarizing accounts, writing sales emails, or summarizing knowledge articles.

While these tools can be leveraged directly on Salesforce data, they can also be connected to Data Cloud to pull in external data that isn’t stored on the core object model (as unstructured text can often be difficult and/or expensive to store).

Say, for instance, a bank wanted to leverage Einstein Copilot to provide an overall summary of each of their accounts so that their bankers would be more able to provide personalized service offerings to them. They could build a custom Copilot action and use Prompt Builder to pull data from Salesforce and a data graph in Data Cloud into a summary, generating a comprehensive view of an account summarized in plain English that can be served up as part of a conversation with Copilot.

Speaking of data graphs, they will help drive customization in Copilot LLM responses, personalization, and more. Data graphs are built from fields selected from a primary data model object (DMO) and its related objects and calculated insights, and can be quickly retrieved in JSON formats. Generative AI models can consume JSON code easily, so data graphs can provide highly valuable inputs to prompts that drive Copilot actions.

If your organization is interested in using your own LLMs to augment Salesforce functionality, you can now set up an external LLM (think about Google Gemini, Meta’s Llama3, or a custom LLM as examples) as a foundation model to serve as a source for generative AI use cases.

Need help getting started with Einstein Studio? Atrium’s experts can help

Implementing new data, analytics, and AI capabilities into your business is no easy task. That’s why we recommend working with an experienced partner. Atrium’s team of data science and AI experts can help you access the insights and solutions you need to achieve your goals.

We’ll help you go from complicated, manual data processes and guesswork to automated recommendations and predictive insights to accelerate your business growth.

Learn more about our data science and AI consulting services.