Workday Rising 2025 - does its acquisition of Sana mean Workday agrees SaaS is dead?

Workday's $1.1 billion acquisition of Sana, announced at the opening of last week's Rising conference, is the third AI startup acquisition by the enterprise applications vendor in the past month, and its fourth in the past year. Workday has been keen to emphasize that this latest deal is not just about enhancing Workday's Learning & Development (L&D) product offering — although Sana will have a big impact here, too — but will also play a huge role in its overall AI strategy.

The phrase that Workday executives have been using over and over throughout Rising has been that Sana will form the basis of Workday becoming "the new front door to work." Not just in Workday's traditional realms of HCM and finance, but across all enterprise functions, and in particular across the entire body of enterprise knowledge and know-how. The underlying thesis is that AI completely upends the way that workers have used enterprise applications to get things done, sweeping away the conventional point-and-click interface to replace this with a conversational experience. In this new AI-led interface, the user just asks questions of an AI assistant or agent and it goes off and fetches an answer or completes an action, or some combination of the two. As Rob Enslin, who as President and Chief Commercial Officer at Workday was closely involved in the Sana acquisition, told diginomica's Jon Reed:

Essentially, we looked at, how do we change this interface in Workday? Because this was the darling of Workday. People love our interface, right? But with AI everywhere, everything we do, every way we did it in the past, it's gone. I do not want to have a pull-down menu and try to figure out, you know, when bonuses get paid in [a] compensation [app]. I just want to ask the question, and the system should tell me.

This sea-change in how workers interact with enterprise apps has been on the cards for some time. Back in 2017, I wrote about the switch to conversational interfaces and the implications for enterprise architectures:

What makes this significant is that we can get a response from the application without ever having to leave the conversational layer — and we can converse with multiple applications all from the same platform. Previously, we had to actually visit each individual application to find information or complete an action. But now all of that workflow can happen in the messaging layer — and the underlying applications become 'headless' as those individual screens and command lines we had to use before now become redundant.

The implication is that the traditional discrete applications we're used to would all retreat into the background, operating behind the scenes, while we the users simply converse with AI agents. At the time, it seemed like this change was just two to three years away, but in tech it always takes longer than you expect for changes of this magnitude to come to fruition. We all tend to underestimate just how much else needs to change in the architecture to make it all possible. I mapped out a concept of a Tierless Architecture that would be needed to enable this change, and in the intervening years many of the necessary elements have been taking shape — in particular a move to a much more composable architectures and more flexible, granular APIs that make it easier to source the precise data or functionality that's needed to answer a question or complete an action. Workday's recent announcements and acquisitions — not just Sana, which in a sense is just the UX icing on the entire tierless cake — are all about building out everything else that's needed to complete the roadmap to reach this ultimate destination.

An AI startup founder mindset

One of the most important milestones on Workday's journey was the acquisition of Evisort a year ago. This brought into Workday the crucial capability of being able to handle unstructured data. This is the kind of data where generative AI really shines, because it shortcuts the previously very laborious process of making human knowledge computer-readable. Instead of having to manually tag the key data hidden away in paperwork, message threads and various other forms of unstructured content created by humans, generative AI — provided it's given enough source material to properly contextualize the content — is able to recognize and categorize that data automatically. Traditional enterprise applications, based on highly structured SQL databases, have never been good at handling this type of data, but with generative AI now able to add the necessary context, it puts a much broader dataset within reach.

But just as important was a different mindset, a fresh view that wasn't encumbered by the historical baggage and assumptions that build up when things have been done a certain way for a number of years. Earlier this summer I spoke to Jerry Ting, Evisort's founding CEO and now also Head of Agentic AI at Workday, who spoke about some of the fundamental changes in how Workday builds and sells products that have been under way at the vendor. Ting isn't the sole architect of these changes, which are being led by Gerrit Kazmaier, President, Product and Technology, who came to Workday in March from Google Cloud, while Peter Bailis, formerly VP of Engineering at Google Cloud, joined as Chief Technology Officer in May. Workday co-founder Aneel Bhusri is also closely involved, bringing his experience of breaking the mold of the previous generation of enterprise software when Workday launched its cloud-native offering 20 years ago. Ting told me that Bhusri wants to recapture that spirit of challenging the status quo:

When I met with Aneel, he told me, 'Jerry, when Workday was 20 people, we thought just like you.' In Aneel's mind and his heart, this is his baby. He doesn't want us to become an old ERP system. He wants us to take the hill.

I met with Aneel as a part of this role, and he asked me, 'Jerry, how do you bring that startup AI founder mindset to Workday?'

Essential building blocks

The past month of acquisitions, along with other in-house development initiatives at Workday, have put further essential building blocks in place. The acquisition of Flowise brought vital low-code agent-building capabilities to the new Workday Build developer platform, which customers and partners can use to build and adapt AI-powered agents and applications. Alongside Build, a significant internal initiative was the creation of Data Cloud, which enables zero-copy access to external data for combination and analysis with data from within Workday's own applications. Unifying the underlying data across previously separate application-centric tiers of data and logic is a necessary complement to the unifying conversational user interface of the Tierless Architecture concept that I mentioned previously. Also important is the introduction of consumption-based pricing, which as Ting explained to me, changes the business model to a mindset that's even more focused on delivering customer value than traditional SaaS.

Last month's acquisition of Paradox brings a conversational experience to the recruitment process. It's also worth mentioning the acquisition of HiredScore earlier last year, which brought predictive AI analysis to the realm of recruitment and internal mobility. Moving to a skills-based talent strategy is another trend that dates back quite a few years, but this too is now accelerating due to advances in technology. It's one area where Sana's technology will have a big impact going forward, as Alan Gray, Senior Principal Functional Architect for Workday Learning, explained in a session at Rising last week:

With AI, if we can continuously assess people, feed that back to the profile and then automatically update as you go forward, then you're in this real area of continuous improvement, continuous performance...

Think of the future where we don't have courses and we don't have programs, we just have content, and based on the needs, the content is dynamically generated for you as an individual to meet your exact need. I think there's some in-between steps to get there. Again, we have some examples of that in terms of compiling learning pathways based on your catalogs and based on questioning and based on some profiling. But potentially that could be there in the future, and it means you could really rationalize or reduce the amount of time people spend taking unnecessary content, but now they have this targeted content, dynamically published.

But Sana is even more important because what it does for learning is actually the same as any knowledge extraction, and therefore the technology is ripe to become the foundation for that "front door to work" that Workday executives have been portraying. As Shane Luke, Head of AI at Workday, told diginomica via email:

By combining Sana's innovative AI-native capabilities with our vast data and contextual understanding, we will be able to deliver a truly intelligent and personalized experience for our customers, where AI agents proactively anticipate user needs, automate complex tasks, and fundamentally transform how people engage with their jobs.

My take

Rising returned to San Francisco this year as part of Workday's celebrations of its 20 years in existence, and the lessons of its rise to prominence at a time when the fundamentals of enterprise application architectures were in flux will not have been lost on Bhusri and others with long enough memories. SaaS vendors now themselves face an existential threat as they adapt to the advent of generative and agentic AI. It's not so much a case of the death of SaaS, because these products are built on a rich catalog of data and know-how, a system of knowledge that AI needs to provide context to its analysis, which means that they're not so easily replaced. But they do face a metamorphosis into a new form, and like the shift to cloud, it's a transition that some vendors will complete in far better shape than others. Workday is taking big steps to ensure that it remains in the vanguard of that transition.

The big change is a shift away from discrete applications with their own separate tiers of UX, logic and data to a unified user experience and an underlying unified data layer, with granular APIs and similar protocols connecting up the business logic. Workday aims to be a player here, too, with its Agent System of Record, which was bolstered last week by the ability to register and manage Microsoft agents.

The speed at which Workday is moving is impressive, but there's still more to do. Jon Reed poses some key questions in his analysis of the past week at Rising, as well as Kazmaier's response to the 'SaaS is dead' line. On my side, I would like to see more of a teamwork element to the Workday vision, which so far still seems rooted in the enterprise and its relationship with individual workers rather than how those workers interact with each other. With agents rapidly changing the nature of everyone's work, how people collaborate to maintain oversight of everything the enterprise aims to achieve becomes even more important.

To answer the question posed in the title of this article, I think that while many of the key foundations on which SaaS has been built will remain, we are approaching the end of the SaaS era and the beginning of the agentic era of enterprise IT. Workday's acquisition of Sana signals its belief that, while not rejecting those SaaS foundations, it's time to move on into this new era.