Main content

Dreamforce 2023 - Salesforce harnesses metadata for its new Einstein 1 platform based on Data Cloud

Phil Wainewright Profile picture for user pwainewright September 12, 2023
A big step for Salesforce as a new Einstein 1 platform based on Data Cloud uses the CRM metadata framework to orchestrate data for use with its workflow, analysis, AI and UX tools.

Patrick Stokes, Salesforce - Zoom screengrab
Patrick Stokes, Salesforce (Zoom screengrab)

Dreamforce rolls around for another year and it's time for a new Salesforce platform. The opening of Dreamforce 2023 today sees the unveiling of the Einstein 1 Platform, which brings together the data connectivity of Data Cloud, the AI capabilities of Einstein, and the metadata framework of the entire Salesforce platform. The new platform unifies the previously separate Marketing Cloud and Commerce Cloud with the existing Sales, Service and Community clouds, at the same time as making it easier to bring external data into the Salesforce environment to be included in AI-powered analysis and workflows.

To entice customers to try out the new platform, from this week any Enterprise or Unlimited edition customer will have free access to Data Cloud for up to 10,000 unified profiles along with two Tableau creator licences.

The enhanced Data Cloud builds on last year's announcement of Genie, which introduced a real-time, hyperscale infrastructure that uses the Salesforce metadata framework to orchestrate data — what I then called a "a new layer of abstraction that normalizes data from multiple sources so that it can be analyzed and acted upon as a single dataset." This is now at the heart of Data Cloud, as Patrick Stokes, EVP of Product and Industries Marketing, explains:

Genie is now Data Cloud. And it is now also on this core metadata framework, which means that we now have a hyperscale data engine directly inside of Salesforce to connect all of your data, and any data that sits in Salesforce is now available as Salesforce metadata, meaning you can use it across any application — sales, service, commerce, marketing, et cetera.

Data orchestration using metadata

Moving to a metadata framework has been the key to bringing marketing and commerce data into the same platform. For many years, these datasets have remained separate from the core Salesforce platform apps of Sales, Service and Community, having retained the pre-acquisition architecture of the original ExactTarget and Demandware products. Merging those data models together with the core Salesforce data model was always too disruptive to contemplate, whereas combining them at the metadata level is a far simpler and less disruptive approach. The same is true when mapping third-party data sources into the Data Cloud.

The metadata approach effectively creates a common language for connecting data across applications, and which is compatible with the many Salesforce tools and functions that have been built to work with that metadata, including tools in Einstein for AI predictions and content generation, the low-code Flow workflow automation tool and the Lightning user interface. All of these functions can access the metadata rather than being directly tied in with integrations to the underlying data. A final benefit of the metadata approach, which Salesforce already takes advantage of in its SaaS applications, is the ability to deliver automated upgrades that won't break automations and extensions that have previously been implemented. Stokes says:

The Data Cloud under the covers is really a very modern data lake house architecture. It's designed for massive-scale data, structured data, semi-structured data, unstructured data. This is unlike anything that you have traditionally been able to put inside or even work with Salesforce with.

Many longtime Salesforce customers will be reasonably familiar with some of the limitations of the standard transactional database that Salesforce has been founded on. What Data Cloud represents is an extension to that. What we've done is we have extended our metadata framework to effectively be able to read and write data from Data Cloud, in the same way that it can our traditional transactional database, which is to say we've decoupled the underlying storage mechanism of Salesforce from its platform.

All of that is a very fancy way to say that you can now have all of this data, either inside of Salesforce or virtually connected via this pioneering zero-ETL mechanism that we've been working on with so many partners to connect data, model it, unify it into a customer record, and then that data is available in all of your Salesforce tools.

If you want to build a business automation with Flow, if you want to build a Lightning page, if you want to connect it to AI in a prompt, it is all available.

From a strategic perspective, this is really about the need to give our customers more capability on this emergence of big-scale data, and to help them go from that big data to really smart data and connect it to the rest of the platform.

The normalized data is also ready to be accessed using the new Copilot UI capabilities announced today, as my colleague Stuart Lauchlan explains in his separate write-up.

My take

What the new Einstein 1 platform offers is essentially a view of your data that is formatted to work with Salesforce's data model and all of its tools for workflow, analysis, AI and the user experience. This will be highly attractive to customers with an existing large commitment to Salesforce, provided they're willing to double down on that commitment and bring even more of their data into the Salesforce metadata layer for data orchestration.

Customers with lesser commitments to Salesforce may feel that they'd prefer to keep their options open, and perhaps use application-level tools from other sources. But that will require either building their own data orchestration layer or using someone else's. They'll either end up equally tied to some other vendor's platform or else spending a lot of technical resource on maintaining their own independent platform.

This evolution of the platform does therefore represent an important strategic step for Salesforce, and one that is based on a significant organic investment in its own technology — the company has assembled a team of very smart data engineers to build many of the ingredients of this new platform. The name change is intended to signal how significant a step forward this is. Longstanding Salesforce watchers will remember that the precursor to the current Lightning UX was originally introduced ten years ago as the Salesforce1 platform. Whether or not the name Einstein 1 is an unconscious or even a knowing echo of that earlier iteration, insiders are talking about this as the biggest evolution of the platform in the past 10-15 years. It's a big move, with important implications for customers.


For all the highlights from Dreamforce 2023, check out our dedicated events hub here.

A grey colored placeholder image