I’ve been on the road for the past five weeks attending technology conferences. The week of September 11, I was in San Francisco for Salesforce’s annual Dreamforce conference. The multiday event is always informative, not only about Salesforce itself but also about the big issues at play for its partner companies and in the tech world at large. Unsurprisingly, this year’s edition put major emphasis on data, AI and trust—all of which I’ll dig into below. AI has been the primary staple food for technology companies since November, when OpenAI and ChatGPT arrived on the scene. I believe this is a trend, not a fad, and will be up there with the web, e-commerce, mobile-local-social and the cloud in terms of enterprise impact.
In his opening keynote, CEO Marc Benioff highlighted Salesforce’s new rank as the third-largest software company in the world (behind Microsoft and Oracle and having just nosed ahead of SAP). He also made a product claim that I have yet to validate, namely that “Salesforce already is the number-one artificial intelligence platform used around the world.” The company also called Dreamforce “the world’s largest AI conference.”
Maybe in keeping with these lofty positions, Dreamforce 2023 introduced a new level of spectacle that I only see these days at events like CES. The crowds. The crush. The giant bears and Einsteins. The lush forestation throughout the Moscone Center . . . it was really something to behold. And it went hand-in-hand with a live music warmup from Dave Matthews and voiceover from Matthew McConaughey. Salesforce gets marketing, and we all saw it in full view.
Integrating data, AI and apps on the Einstein 1 Platform
Amid all the hype, the clear star of the show was the Einstein 1 Platform, which pulls together Salesforce’s data cloud, CRM applications and predictive and generative AI (GAI) capabilities inside a single metadata framework. It builds on the company’s Einstein AI, which was launched in 2016, and earlier AI research going back to 2014.
The idea driving Einstein 1 is compelling. First, create a single repository of truth about customers that can be accessed by Salesforce Sales Cloud, Marketing Cloud, Commerce Cloud and so on. Use an open framework so it can interact with a broad range of other software, including in-house apps such as Slack and Tableau or external ones like Google Workspace and Microsoft 365. Then, turbocharge your employees’ productivity by providing them with AI-driven “copilots” that operate seamlessly anywhere across the Salesforce landscape. Even better, make it extremely simple for non-programmers to customize their workflows with low-code and no-code functionality.
There’s a lot to like here. During the keynote session, Benioff emphasized how important it is to create an integrated data platform to overcome the status quo that too many companies live with, in which having disconnected data, apps, APIs and vendors leads to low productivity. The market needs something different, he said, before it wades into full-bore use of GAI, autonomous AI-driven agents and the like. “We have found the metadata framework is the key to integration,” he added—and I see his point. Salesforce has already integrated a ton of valuable assets within that framework, and over time it will continue to expand its usefulness by bringing even more assets into the fold.
What about all the other data out there?
So far, so good. But I do have some reservations. For one thing, it’s great if a company’s using a data lake from Snowflake, AWS or whoever to house its data; clearly, you can hook that right up to Einstein 1, and it will be cleaned, harmonized and put to use right away. But what about the 75% to 90% of all enterprise data that’s still locked inside on-premises systems? In another part of the keynote session, Salesforce CTO and cofounder Parker Harris noted that 25 years ago, people didn’t trust having their data on the internet at all. I wanted to say, “And most of them still don’t.”
What is Salesforce doing to get all of that data onto the Einstein 1 platform? Sure, what’s already available for use by Einstein 1 is great—but does the rest of that data matter? While I’m at it, Cloudera has the largest amount of on-prem data under management of any company. Why isn’t it a Salesforce partner?
This leads to broader questions: Why aren’t the leading ERP companies—especially Oracle and SAP—partners, either? And how will Salesforce apply GAI in a truly cross-enterprise way without ERP data?
Salesforce’s answer to integrate that data is Mulesoft. If your favorite data isn’t with Salesforce’s partners or AppExchange, you’re left with doing some serious data integration with Mulesoft. While this isn’t a showstopper, it does need to be recognized as a heavy lift to integrate SAP, Oracle and IBM data.
Let me put this in the context of the customer example that Salesforce used during the keynote session. Benioff did a little interview right there on the ballroom floor with Laura Alber, who’s the CEO of Williams-Sonoma—which uses Salesforce extensively. It was a nice touch, especially since the audience had just seen a video about the retailer that featured a smart time-lapse storyline about a (fictional) customer family that might have been taken from Shrinking or This Is Us. Demos from the stage focused on the same longtime Williams-Sonoma customer from the video—a woman who was now awaiting a shipment of chairs for the new restaurant she was about to open.
The demos showed how a marketer could use the AI functionality of Einstein 1 to correctly reassign this customer from a loyal-consumer segment to a B2B segment, and how a support rep could come up with a better product option when the desired chairs wouldn’t be available in time for an earlier opening date. And a sales rep could set up a meeting automatically or generate quotes, proposals and even contract language on the fly.
As you can guess, the demos were all impressive, and it’s not hard to imagine plenty of examples of improved productivity and problem-solving enabled by Einstein 1. But what if all the supply chain data for the Salesforce customer in question lived in an on-prem system? What if the problem at hand required access to HR information, or the latest data from manufacturing? How about procurement, product development, legal, finance?
Kudos to Salesforce for breaking down so many data silos. But Einstein 1 will need to bring in even more kinds of data to truly fulfill its potential.
Bridging the “AI trust gap”
One challenge that the Salesforce executives took head-on during the keynote was the question of trust. Benioff put it simply when he said, “We want to build at Salesforce the trusted AI platform for customer companies.” That’s a tough task when 52% of consumers don’t believe that AI is safe and secure, according to a Mitre-Harris poll cited during the presentation.
Benioff outlined some of the trust challenges related to AI, including LLMs stealing data, GAI “hallucinations” and the toxicity and bias that can arise if AI models are not carefully tuned. All of this will become even more challenging as we move from the first two “waves” of AI—predictive and generative—into ubiquitous autonomous AI for self-driving cars and the like, and finally into artificial general intelligence (AGI, which is what movies like The Terminator exaggerate for dramatic effect).
One major way Salesforce is addressing this issue is with the Einstein Trust Layer. Harris went through this feature in some detail, and it’s clear that Salesforce has put a lot of thought into how it can harness the power of LLMs—both its own and others from the likes of OpenAI and Anthropic—without putting sensitive enterprise data at risk. The Trust Layer includes automated steps to check for toxicity and hallucinations, plus it generates an audit trail so users can see where GAI results come from.
Creating the most trustworthy AI platform is a great goal, but how does Salesforce intend to nail down that attribute—to make it tangible for customers and prospects? During the keynote, Benioff’s slides showed off some of the “citizenship” accolades his company has earned over the years, including being named to lists of the most sustainable and most ethical companies. Shouldn’t there be a score that helps determine the same thing for AI trustworthiness? If not, claiming to be the “most trusted” AI company comes across as a throwaway line. I recommend that Salesforce create its own metric for this, or else leverage one created elsewhere that could serve as an industry standard.
Plenty of promise, but also missed opportunities
In the understatement of the day, Benioff said, “There is no question this AI opportunity is going to change everything.” No question about that, indeed. President and chief product officer David Schmaier supplied lots of specific examples of what Einstein 1 can change when he gave a quick tour of the copilot features in everything from Marketing Cloud (“It’s mass personalization, at scale, for your entire marketing department”) to Tableau (“Now everyone is a data analyst”).
As I said above, it’s compelling stuff. However, it seems to me that Salesforce is using Einstein 1 to target its current customer base to get them to buy more, versus using it to attract new customers. That approach has merit, especially when you consider that Salesforce’s market share in CRM is already as large as its next few competitors’ put together. (Salesforce is at 23%, and no one else even cracks 6%.) Einstein 1 is also likely to have the greatest appeal for companies that are the most open to using their enterprise data in the cloud, which by definition includes every Salesforce customer.
That said, the market opportunity might eventually grow even larger if Salesforce can appeal to companies that haven’t made big moves to the cloud yet. Because if the company can persuade new customers to move even a small fraction of that 75% to 90% of on-prem enterprise data to the cloud . . . the sky really is the limit.
Out of the other side of my mouth, I will say that Salesforce has the highest short-term generative AI opportunity with its installed base before it goes after new customers.
One more thing, which I’ve saved for last because Salesforce tucked it away at the end of the keynote session: Where is Slack in all of this? Slack stands out to me as the ultimate copilot—I mean for the way people already use it in their daily workflows, and for how it could be used within the Salesforce ecosystem.
After a string of well-executed presentations from Salesforce leaders and technical experts, Slack CEO Lidiane Jones took the floor and gave the audience an extra jolt of energy with a compelling demo of the new cross-functional and AI-driven features in Slack. Talk about enhancing productivity—how about coming back to Slack after a day of client meetings and having it catch you up with quick summaries of team activity, milestones achieved and next actions? The functionality should be especially useful for sales teams, and Jones said that her goal is to save every seller one day per week through these enhancements.
But if all of Einstein 1 is driven by chat-based, natural-language interactions, why wasn’t Slack—the greatest chat-based tool in the corporate world—used as the interaction mechanism for the new platform? For that matter, why didn’t Salesforce use at least the Slack brand, which is already synonymous with getting instantaneous chat-based answers?
Mark that one down as a missed opportunity. Now it will be fascinating to see what Salesforce is able to do with the wealth of other opportunities that Einstein 1 will create.
Follow me on Twitter or LinkedIn. Check out my website or some of my other work here.
Moor Insights & Strategy provides or has provided paid (wish services to technology companies, like all tech industry research and analyst firms. These services include research, analysis, advising, consulting, benchmarking, acquisition matchmaking, and video and speaking sponsorships. The company has had or currently has paid business relationships with 8×8, Accenture, A10 Networks, Adobe, Advanced Micro Devices, Amazon, Amazon Web Services, Ambient Scientific, Ampere Computing, Analog Devices, Anuta Networks, Applied Brain Research, Applied Micro, Apstra, Arm, Aruba Networks (now HPE), Atom Computing, AT&T, Aura, Avaya Holdings, Automation Anywhere, AWS, A-10 Strategies, Bitfusion, Blaize, Box, Broadcom, C3.AI, Calix, Cadence Systems, Campfire, Cisco Systems, Clear Software, Cloudera, Clumio, Cohesity, Cognitive Systems, CompuCom, Cradlepoint, CyberArk, Dell, Dell EMC, Dell Technologies, Diablo Technologies, Dialogue Group, Digital Optics, Dreamium Labs, D-Wave, Echelon, Elastic, Ericsson, Extreme Networks, Five9, Flex, Fortinet, Foundries.io, Foxconn, Frame (now VMware), Frore Systems, Fujitsu, Gen Z Consortium, Glue Networks, GlobalFoundries, Revolve (now Google), Google Cloud, Graphcore, Groq, Hiregenics, Hotwire Global, HP Inc., Hewlett Packard Enterprise, Honeywell, Huawei Technologies, HYCU, IBM, Infinidat, Infoblox, Infosys, Inseego, IonQ, IonVR, Inseego, Infosys, Infiot, Intel, Interdigital, Intuit, Iron Mountain, Jabil Circuit, Juniper Networks, Keysight, Konica Minolta, Lattice Semiconductor, Lenovo, Linux Foundation, Lightbits Labs, LogicMonitor, LoRa Alliance, Luminar, MapBox, Marvell Technology, Mavenir, Marseille Inc, Mayfair Equity, MemryX, Meraki (Cisco), Merck KGaA, Mesophere, Micron Technology, Microsoft, MiTEL, Mojo Networks, MongoDB, Movandi, Multefire Alliance, National Instruments, Neat, NetApp, Netskope, Nightwatch, NOKIA, Nortek, Novumind, NTT, NVIDIA, Nutanix, Nuvia (now Qualcomm), NXP, onsemi, ONUG, OpenStack Foundation, Oracle, Palo Alto Networks, Panasas, Peraso, Pexip, Pixelworks, Plume Design, PlusAI, Poly (formerly Plantronics), Portworx, Pure Storage, Qualcomm, Quantinuum, Rackspace, Rambus, Rayvolt E-Bikes, Red Hat, Renesas, Residio, Rigetti Computing, Ring Central, Salseforce.com, Samsung Electronics, Samsung Semi, SAP, SAS, Scale Computing, Schneider Electric, SiFive, Silver Peak (now Aruba-HPE), SkyWorks, SONY Optical Storage, Splunk, Springpath (now Cisco), Spirent, Splunk, Sprint (now T-Mobile), Stratus Technologies, Symantec, Synaptics, Syniverse, Synopsys, Tanium, Telesign,TE Connectivity, TensTorrent, Tobii Technology, Teradata,T-Mobile, Treasure Data, Twitter, Unity Technologies, UiPath, Verizon Communications, VAST Data, Veeam, Ventana Micro Systems, Vidyo, Volumez, VMware, Wave Computing, Wells Fargo, Wellsmith, Xilinx, Zayo, Zebra, Zededa, Zendesk, Zoho, Zoom, and Zscaler.