While many organizations are still building out their data platforms, Salesforce Inc. has taken a monumental stride forward. By seamlessly incorporating metadata integration, the company has transformed the modern data stack into a comprehensive application platform known as the Einstein 1 Platform. 

Under the leadership of Muralidhar Krishnaprasad, executive vice president of engineering at Salesforce, the platform — constructed on the company’s underlying metadata framework — harmonizes metadata and integrates it with AI and automation, ushering in a new era of data utilization.

What’s behind the Einstein 1 Platform and its innovations to the Salesforce Data Cloud, as well as Einstein 1’s AI capabilities? And what does this integrated platform accomplish for organizations? 

The company set out to create a platform that empowers all business users — salespeople, service engineers, marketers and analysts — to access, use and act on all their data, regardless of its location, according to Krishnaprasad. Salesforce’s open, extensible platform not only allows organizations to unlock their trapped data, but also equips them with generative AI functionality, enabling next-level personalized experiences for employees and customers.  

“Analytics is very important to know how your business is doing, but you also want to … make sure all that data and insights are actionable,” Krishnaprasad said. “That’s our big thing — to make sure we blend all three together: AI, automation and analytics. It’s really important that you also bring the metadata layer …  that’s the big secret sauce that we brought together.” 

Krishnaprasad spoke with George Gilbert, senior analyst at theCUBE Research, in the latest episode of “The Road to Intelligent Data Apps” podcast series, theCUBE’s ongoing exploration into the frontier of intelligent data applications. They delved into metadata integration, the open-API technology behind Einstein 1, the platform’s key features and how its extensibility and interoperability enhance its usability across a variety of data formats and sources.

Metadata integration: A diverse tech stack built to accommodate virtually any IT environment

Built on Trino, the federated open-source query engine, and Spark for data processing, the platform’s rich set of connectors and open, extensible environment builds in a wide array of options. Organizations can freely share data between warehouses and lake houses, among other systems. But that’s just for starters.

“We also use this thing called a hyper-engine … the in-memory overlap engine we use in Tableau,” Krishnaprasad said. “If you’re doing Tableau or other data exploration, you want that sub-second response time. We automatically do all those internally.”

The platform accommodates a wide range of machine learning options. Users can also add their own large language models to the stack. Whether they use the Salesforce Einstein integrated platform, Databricks, Vertex, SageMaker or other solutions, they can do so without  needing to copy any data, according to Krishnaprasad. 

“You can literally point the tool, create the models and then score any data in there. We also took [that same flexibility] to the LLM,” he said.

The platform includes three levels of extensibility. This feature allows organizations to standardize and extend their customer journey models easily, Krishnaprasad pointed out. The platform starts with basic reference models, such as customer names, which can be customized by adding new fields or creating new models and relationships. On the next level, these models can be utilized to generate various insights, including business intelligence or AI-driven insights. Finally, at the top level, users can introduce their own functions or triggers to act on these insights.

The platform also provides considerable flexibility in the unification process. For instance, while many systems may only perform unification once, the Einstein 1 Platform continuously performs that function, and users can also create different unified graphs. 

“Remember, we’re a multimodal system, meaning we’re looking at your entire customer journey,” Krishnaprasad said. “A marketing team might decide that a 60% match may be OK if they’re doing advertising. Whereas if you’re servicing somebody or you’re a bank or you want to do a financial thing, you want, probably, a highly deterministic match. We allow you to create different graphs, fully controllable by the customer. We give you flexibility at almost all levels of the stack for you to be able to create that right experience for your business.”

The triad that powers a transformed customer journey: AI, automation and analytics

Whether batch, streaming or real time, the platform’s foundation progressively ingests, harmonizes and unifies data. This functionality results in a standardized metadata model that creates, extends and enriches a single 360-degree view of what customers are doing with your business. 

Achieving this view is a game-changer for organizations because it frees siloed data — the vast majority of which is stuck in unstructured forms, according to Krishnaprasad. This data — which might include conversations, Word docs, PDFs, emails, audio, video and more — is a goldmine, and the platform lets customers extract and process it.

“What we have done with this customer 360-degree model is to be able to use that unified data, from wherever your data may be, generate the insights on top of the data and make this … unified data and insights accessible on all of these application surfaces — and, in many cases, even push actions so you can go react to these things,” Krishnaprasad said. “That’s really the bigger customer journey we’ve unlocked.” 

For example, when a customer views an ad and then visits the website, the company will know who they are. Salespeople know precisely what site visitors are interested in. Service personnel understand what customers are unhappy about. Analysts can easily access all the information they need to generate business-enhancing insights. Through these capabilities, customer engagement soars, according to Krishnaprasad.   

“Couple that with generative AI and we can take it to the next level, because then you can even start doing a lot of self-service around these things,” he said. “You want to make sure you’re giving the right answers, and that’s what we’re enabling with our data platform. We are elevating that data to a higher level to be able to create that unified model and then … power a unified experience across your entire customer journey.”

Here is the complete conversation, part of the “Road to Intelligent Data Apps” series: