The ultimate goal of pulling customer data together into a customer data platform (CDP) is building more meaningful customer experiences in real time. Up until now, that’s been more aspirational than real, but Salesforce is announcing Genie, a real-time data integration platform, today at the Dreamforce customer conference, which aims to make that dream a reality.

At its core, Genie is a new data integration model that underlies the entire Salesforce platform with the aim of moving data wherever it’s needed most — and doing it fast.

Patrick Stokes, EVP and GM of platform at Salesforce, says this is probably the biggest news coming out of Dreamforce this week. “Genie effectively enables the world’s first real-time CRM,” he said.

“So we’re announcing that our Customer 360 applications — sales, service, commerce, marketing, everything in our Customer 360 portfolio — now have access to an entirely new way of bringing data into Salesforce in real time at scale that we’ve never been able to achieve before. And with that, our users can orchestrate real-time customer experiences against those datasets,” Stokes explained.

Prior to this, the company had built data integrations based on the transactional data in the Salesforce CRM database. This goes back to 2007 when Salesforce announced plans for Force.com at that year’s Dreamforce. Stokes said Genie is the modern equivalent of that early attempt, using a data lake that the company built to store the data instead of a transactional database.

“We connected this lakehouse architecture to the Salesforce platform, which at the technical layer means literally, we taught it Salesforce metadata, which is the way that all of our services talk to each other.” This approach also allows the platform to work with external services and data repositories, as well. In fact, the Snowflake integration the company announced last week is built with this technology.

But Genie is more than just a data integration layer. By allowing data to flow faster and more freely, it opens up all kinds of automation possibilities, especially when you combine it with Einstein for AI and machine learning and Salesforce Flow, the company’s workflow tool.

“If your platform can suddenly talk to all of this new data, and that data is coming in in real time, then you can use our automation layer like Salesforce Flow to orchestrate workflows or automations in real time, but only if the platform can keep up with the speed of change and volume of data that’s coming in,” he said.

Part of the ability to go faster beyond the architectural changes at the software level is that Genie is running on Salesforce’s own cloud infrastructure, Hyperforce, which was announced in 2020 as a way to move data from Salesforce to the public cloud. In this case, they are using it to move data between Salesforce and other services, both on the platform and to other data sources like Snowflake or Amazon SageMaker.

He adds that this ability to move data around in real time (or near real time), creates what is essentially a customer data graph.

“When you connect all of these different data sources into Genie, be those directly or other data lakes like Snowflake, what you’re doing is you’re modeling the data. You’re basically hooking it up to a data model. And when you do that, you’re creating a graph of how all that data is related to each other, independent of where it lives in a particular system of record, which is incredibly powerful,” Stokes said.

Liz Miller, an analyst at Constellation Research, says the shift to a new data model is a much-needed move for the company by pushing the CDP beyond marketing

“Honestly the thing I find most important about this is that Salesforce is moving in the right direction with their vision of a customer data platform. They are not treating a CDP as if it is a marketing toy for marketing things. Instead, they are turning the CDP into a foundational layer of unified, normalized and persistent personalization and smart segmentation that benefits the entire customer experience front line across sales, service and marketing,” Miller told TechCrunch.

Sheryl Kingstone, an analyst at S&P Global Market Intelligence, who has been covering the CRM space for years, agrees, saying the key to this change is building the data mechanism in a way that you can share this valuable data more widely.

“They are really focused on building this as part of what I would say is a true platform with all of the assets that this needs to work, and hopefully, it will create what I call a ‘customer intelligence platform,’ which makes sure that you don’t have multiple different CDP silos. And we finally can have that single source of the truth and execute on it.”

The combination of tooling has the potential to be able to make things happen based on the data and the situation without requiring human intervention, and that can be powerful. But Kingstone says the human side still matters and companies have to learn to put data in the hands of the people on the ground working with customers.

That’s going to be a huge challenge, regardless of how sophisticated the technology is, but Salesforce is attempting something big here that’s never been done before by changing the way data moves around the platform. Whether that truly leads to better customer experiences, online and in person, however, remains an open question.

Unlike many Dreamforce announcements, customers don’t have to wait until next year for Genie. These new capabilities are available now.

• Salesforce's widespread adoption makes its data critically important to operations.
• But many tools, like Snowflake, have emerged to make Salesforce data points one of many in analytics.
• Salesforce is launching a new tool to power real-time updates and bring usage back to its platform.

As tens of thousands of Salesforce partners, developers, and customers descend on its Dreamforce conference in San Francisco this year, the software giant is finding itself increasingly intertwined with another unexpected data giant: Snowflake.

Increasingly, analysis of the data generated via Salesforce is not happening on Salesforce. It's instead getting piped into Snowflake or other providers, where the work — such as building machine learning models or analyzing customer demographics and behaviors — actually happens alongside other data sources like those from Twilio's Segment. And Salesforce is addressing that uncomfortable reality with several big strategic moves.

"Set simply, the pattern we see customers doing, they need to bring in data from a lot of different sources — Salesforce included — and they need to do work," Salesforce EVP and GM of platform products Patrick Stokes told Insider in a recent interview. "They need to build an AI model or rationalize the data in some meaningful way so they can derive some insight. Historically, Salesforce has not always been the place to do that."

First, it launched a new tool called Genie in Tuesday at its first conference since the pandemic brought in-person mega-events to a screeching halt. Genie enables customers to orchestrate a variety of tasks and update their products instantly. That can come in the form of recommendations, fraud detection, or others. Genie, too, integrates directly with Tableau, providing a way to create real-time visualizations without having to build them on top of data in Snowflake.

It also said in early September it was partnering with Snowflake, making it natively compatible and integrating it without users needing to copy data. And Salesforce on Tuesday said it is launching a direct integration with Amazon's machine learning tool SageMaker to support custom machine learning models.

At its Tableau keynote, Salesforce emphasized the company's Einstein integrations with Snowflake, with the presentation working with Snowflake data. And at a separate meeting with reporters and analysts Tuesday, co-CEO Marc Benioff took a moment to lean over and remind his co-CEO Bret Taylor to talk about its data lakehouse architecture—and how it plays nice with Snowflake.

"Snowflake has been a close partner for a while, we're going to try to be compatible for all," Taylor told Insider. "The principle of Genie is, these data lakes are huge, and customers don't want to copy data back and forth. We're trying to do zero copy architecture with partners so you can have better governance and lineage, and the chief compliance officer can wrangle that data lineage."

In short, Snowflake was everywhere. And with these moves, Salesforce is ensuring the analysis and usage of Salesforce data and its results can remain inside Salesforce's sphere of influence. 

Salesforce has for the majority of its twenty-plus years of operating enjoyed both carving out and owning a large share of online enterprise tooling. Now, as dozens of startups remake the analytics ecosystem and companies collect more and more data, Salesforce has to adjust.

Salesforce now also joins several companies, including Databricks and Snowflake, as it looks to inch the world closer to a future where products and AI models update instantaneously. Investing in those instant updates can lead to outsized returns, such as a smarter product recommendation that can help close a sale. 

"It really starts to show Salesforce isn't looking at products, they're not looking at individual clouds, they're saying 'how do we be that business operations platform that has to deal with engagement or personalization,'" said Liz Miller, vice president and principal analyst at Constellation. "Snowflake isn't going away—it's not that you suddenly have this sigh of relief and  you're like, 'we dont need Snowflake.' This is more about accessibility and being able to allow your teams to have that real-time access you need."

Snowflake declined to comment on this story.

How Salesforce risked losing control of the usage of its own data

Salesforce operates what it calls a customer data platform, based on a "data lakehouse" architecture built on the Apache Iceberg software. It enables organizations to connect other data lake tools and operate Salesforce data. Snowflake recently backed Apache Iceberg as its file format for its data lake tools, which themselves compete with Databricks' Delta Lake.

But in multiple conversations, industry insiders have been increasingly concerned about whether Salesforce users will simply do all their data analysis through platforms like Snowflake, with Salesforce's tools insufficient for the scale of data operations today. Instead, Salesforce data is part of a conglomerate of sources that land in a data warehouse like Snowflake through popular data analytics tools like Fivetran and Dbt. That data can then go into other sources like Google Cloud's Looker, or various machine learning tools.

From there, analysts and data scientists use other tools to do their work, and pipe the results back into Salesforce. The emergence of tools in the "reverse ETL (extract-transform-load)" space like Hightouch and Census have further trivialized this process and built substantial businesses.

While it isn't necessarily bad for Salesforce that customers are using data from the platform in other places, it also opens up entry points for other companies to disrupt Salesforce's other products — such as Tableau, the visualization tool it bought for $15.7 billion in 2019. Tableau is up against stiff competition, including from Looker.

Salesforce, however, has no intention of building its own data warehouse, Stokes said.

"We're not skipping the data warehouse, we're embracing it," Stokes said. "We're using that capability and integrating it with the Salesforce platform."

Salesforce is now one of many chasing a real-time future

The explosion in machine learning tools that power products ranging from ecommerce to financial services has increased the demand for tools that power real-time data pipelines. But streaming for a long time felt like it was "coming next year" until 2022 turned out to be that "next year."

"I think a big part of what we were seeing is some really scrappy cloud-native businesses start to achieve some degree of real-time within their customer experience," Stokes said. "As some of those companies started to achieve certain parts of the experience in real-time, that changes consumer expectations everywhere."

As a result, Snowflake, Databricks, Confluent, and many other companies are chasing a world where all data is processed and updated in real time. That's in lieu of existing processes where companies update their tools with large batches of data, sometimes once or twice a day. Increasingly, the AI industry is moving to a "micro-batch" approach, with real-time updates as the natural next step.

"Streaming introduces a different velocity of data, and the way to make decisions that take advantage of that unique new frontier of data velocity is one that is not conducive to humans making these decisions," said Mike Del Balso, co-founder of $900 million real-time machine learning startup Tecton. "You can't have a human there making these decisions every time."

While Salesforce said it is largely targeting broad adoption with low-code tools that require a minimum of programming expertise, it will increasingly build connectors that will enable companies to deploy their own machine learning models may be built with outside tools like Hugging Face. 

Salesforce and WhatsApp have entered into a new partnership that will bring instant messaging facilities to the Customer 360 platform.

The partnership, announced during Dreamforce 2022, will deliver businesses a new way to communicate with customers, courtesy of WhatsApp integrations for multiple Salesforce CRM applications.