I recently attended the Salesforce Trailblazer DX event to learn more about Salesforce’s artificial intelligence products and strategy. Fueled by generative AI, awareness and investment in AI seems to be exploding. ISG research shows that enterprises plan to nearly triple the portion of budgets allocated to AI over the next two years. This doesn’t come as a big surprise when you look at the outcomes enterprises are achieving: Of those that have invested in AI, more than 8 in 10 (84%) have had positive outcomes.
However, AI is not without its challenges. A lack of skills is the top issue organizations face. AI-related skills are the most difficult technical skills to hire and retain. There are also issues with trust and accuracy, among others.
The Trailblazer event is oriented more towards developers compared with the company’s flagship event, Dreamforce. And you will see that new capabilities are geared toward those building AI but with the objective of making more AI functionality available to end users of Salesforce applications. Salesforce’s Einstein represents a different, innovative approach to AI. Shortly after Einstein was introduced, we recognized it with our 2018 Digital Innovation Award.
Einstein is an AI platform that is tightly coupled with the various applications Salesforce offers. The advantage of this approach is that it gets AI into the hands of line-of-business workers, where it can have the most impact. The disadvantage is that if you want to use it elsewhere, you will be disappointed.
Einstein 1 Studio is a set of low-code/no-code tools that can configure and customize Einstein Copilot as well as the prompts and actions associated with it. The tools enable developers to work with both generative AI and predictive AI. Developers can choose which large language models to work with and import predictive models developed in other platforms. Prompt Builder enables the creation of reusable prompts to make it easier for end users to quickly generate responses they know are reliable. The product also includes prerelease capabilities to create prompt templates so a single prompt can be used across a variety of similar situations. Copilot Builder and Copilot Actions, both prerelease, support configuration and customization of Einstein Copilot. Developers can tie in other Salesforce tools, such as Apex, Flow and Mulesoft, to coordinate and complete tasks based on interactions with the Copilot.
A number of significant enhancements to Salesforce Data Cloud are coupled with the Einstein functionality. Salesforce recognizes the need to have trusted data−often external to Salesforce−to support AI initiatives. As a result, it has invested in Data Cloud to provide an extensible customer data platform that meets those needs. As my colleague Matt Aslett explains, one of the ways to improve trust in generative AI is with vector search and retrieval-augmented generation. Data Cloud now includes a preview of vector database and vector search capabilities. It also includes a preview of support for unstructured content, an important element in generative AI implementations. Enterprises can now bring their data lake into the data cloud environment. The data does not have to be moved or copied; it can be accessed in place and treated like other Data Cloud data.
Einstein 1 Studio and Data Cloud are certainly capable products that could potentially stand on their own. That’s not the way they are being delivered. These products are for use within the Salesforce ecosystem. Salesforce customers should be pleased with the new functionality, which is easily consumed from within the applications. The challenge will be for enterprises looking to combine AI efforts across the entire information estate. The new capabilities provide mechanisms to bring in external data and models, but there will still be a divide. Nonetheless, the capabilities of Einstein 1 Studio are compelling, and I encourage Salesforce customers to evaluate how to use them to enhance the value of applications with AI.
Regards,
David Menninger