Salesforce is all in on ‘unstructured data’

by:
last updated:
einstein ai

Salesforce is all in on ‘unstructured data’

by:
last updated:

Hat’s off to Salesforce for jumping into the wild and woolly world of unstructured data; whatever happens, it’s going to be a wild ride ahead.

It was 25 years ago today that Mark Benioff told the band to play, and boy, has Salesforce come a long way since then to become one of the leading enterprise software companies in the world. From initially focusing on CRM (though that remains central), over the years, Salesforce has expanded its reach to Marketing, Commerce, Service, etc. In the past year or so, it has focused ever more on its Data Cloud offering and how that is tied to its Einstein AI products.

Einstein, the focus of this year’s TrailblazerDX event, has had quite a journey since its initial launch in 2016. And it’s worth remembering that Data Cloud was only really launched as Genie at Dreamforce 2022 (and then renamed as Data Cloud in early 2023), so it too is still young and something that the developer community is still getting to grips with. With all that being said, this year’s event, thematically at least, revolved around the launch of Einstein 1 Studio, a broad set of tools that provide developers with all they need to customize and build on Salesforce Einstein Copilot, its generative AI assistant. But beyond the big announcements, the hordes of white-coated, white-wigged Einstein lookalikes (my Uber driver thought she was picking me up from a medical conference), and the flashy gold hoodies of the Trailblazer elite, there was another, equally, if not bigger story emerging, that Salesforce is now all in on unstructured data. 

For a few years, Salesforce has banged the drum around the mantra “CRM, AI, Data & Trust,but the data element has always been the structured, not the unstructured data. This makes perfect sense as the vast majority of Salesforce work, particularly in CRM, is by default transactional, and therefore, its focus has been on traditional, structured data sitting in neat rows and columns in a database. However, the emergence of generative AI provides an opportunity and a considerable challenge to tap into the world of unstructured data. To put this into context, I have been covering Salesforce on and off from its earliest days, and until this event, ‘unstructured data’ was a term rarely uttered, as it represented to Salesforce the dark side, the twilight zone of computing, a place best avoided. The executives I spoke to at Salesforce this past week seem to understand that they must tread carefully.

As I mentioned, the focus of the event was Einstein AI. Plenty of content and information was supplied regarding the developer studio (you can read more about that in Matt’s blog here). Still, there was also, in my conversations at least, a lot of talk about vector-based search, data graphs, IDP, and RPA. To put it another way, the influence of the MuleSoft acquisition and its potential importance to Salesforce is finally being felt. All these announcements and conversations boil down to the fact that beyond core CRM, Salesforce is slowly pivoting to becoming the platform for business applications it has long promised, which has enormous implications for the developer community. Indeed, in the long term, this embrace of automation and unstructured data can provide the means to reinvent and take Salesforce to its 50th anniversary. Even so, to quote one Salesforce exec, “This is just the first innings,” the challenge in 2024 is identifying where to start, where the quick wins are, and where other more significant opportunities lie; that will take a lot of time and work to reach them.

The first step is to enhance and extend the core CRM, making unstructured files more accessible, pulling data from them to support transactions, and building simple but critical workflows. That might not sound ambitious, but it makes perfect sense, as those activities open up many short-term developer options. When you mix in the potential of AI, it’s transformative.  But back to my cautionary stance on unstructured data, as any subscriber or reader of Deep Analysis research knows, unstructured data is a messy world to be involved in, far more complex and fuller of traps than most realize. That Salesforce is now ready to embrace it is a good sign. Still, in our estimation, it should first focus on transactional opportunities, build out its RPA and IDP offerings, and get its ecosystem and developer community up to speed and skilled as quickly as possible.  Transactional opportunities in the sense of where unpacking and leveraging a document or file in an existing, traditional Salesforce workflow will make good sense. Where it needs to tread far more cautiously is in tapping into unstructured data silos, Salesforce has the connectors, but connections (APIs) are the easy part; much harder, and, at times, near impossible is to decipher, organize, clean and leverage the contents of those silos. And as many others have learned to their cost over the years, it is often not worth the time, money, and effort to do so. In summary, Salesforce is now looking to execute on a broader enterprise base than before. It has to take on unstructured business data to do that, but it must carefully choose its opportunities in partnership with its developer community.

If ‘unstructured data’ was the mantra at this year’s TrailblazerDX, next year’s event will be even more interesting. Progress over the coming year will depend on how the developer community embraces and learns complex new skills well outside their comfort zone and expands its repertoire to address unstructured data opportunities. Hat’s off to Salesforce for jumping into the wild and woolly world of unstructured data; whatever happens, it’s going to be a wild ride ahead, but Salesforce has something many of its competitors do not: a truly dedicated and enthusiastic developer community that will likely be up for the challenge.

Subscribe to our monthly newsletter

Leave a Comment

Subscribe to our monthly newsletter

We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it.

Work Intelligence Market Analysis 2024-2029