It’s probably not that controversial to say that things are changing quickly. The promise (or the threat) of artificial intelligence is roiling the media, upending traditional ways of doing business, and changing people’s lives. So the language of the software world is changing, too.
Case in point: In the cloud era, people referred to “software as a service” to talk about the elastic and on-demand nature of software applications, delivered through the cloud, with subscriptions. That replaced the old “out of the box” model for software, which you may recall walking through that specific aisle of Staples or OfficeMax.
Software as a Service or “SaaS” revolutionized how businesses access software services. The emergence of the cloud as a connectivity principle spawned all kinds of other similar acronyms, too, among them: Infrastructure as a Service (IaaS), Platform as a Service (PaaS), etc. (Forbes Technology Council member Kalyan Kumar mentions another one, XaaS or “anything as a service,” then contrasting this to the new SaS model.)
Now, here’s where things, as they say, get funny….
We’re in the era of agentic AI, and so, things are different. If you haven’t heard the phrase “Services as Software,” that’s partly because it hasn’t been around long. Like the term “vibecoding,” attributed to Andrej Karpathy, “Services as Software” has a defined origin, attributed to Phil Fersht of HFS Research, who apparently was “Fersht” to coin this term, and apparently, trademark it. (Sorry, I couldn’t resist.)
Ok, so what is Services as Software, and perhaps more importantly, what do you do with it?
Integrating Services as Software Into Business
Essentially, Services as Software means that “services” traditionally provided by humans are made into “software,” albeit software run by thinking LLMs that are doing what they’ve demonstrated their ability to do – run tasks autonomously.
So things like call center work, tax prep, letter writing by paralegals – those things that are called out as white-collar work ripe for automation – will be made into Services as Software.
Here’s how this is described over at HFS:
“Many people-based services, once they become predictable and routine, inevitably evolve into automated processes. With the growing sophistication of AI tools, service efficiency and personalization are reaching unprecedented heights.”
Speed, cost effectiveness, and scalability are touted as benefits of what the writer abbreviates as SaS. (I’ve seen some abbreviate Services-as-Software as “SaaS” which I think is unhelpful and confusing.)
“As the lines between service and software blur, businesses can now access pre-built solutions, automated workflows, and data-driven insights,” HFS analysts write. “This creates a seamless experience, empowering companies to scale faster, adapt quickly, and achieve more impactful outcomes.”
That idea of blurred lines, I think, is relevant: We’ll see more and more of what humans used to do augmented, enhanced, and then possibly fully taken over, by software tied to evolving neural nets. So those “services” are highly vulnerable to automation.
You can also watch a video of someone from HFS going over this concept at a recent Stanford event here…
Smarter Systems
There are numerous ways to describe this change: SaS is but one. Take a look at how Falk Gottlob writes about this on Medium, contrasting yesterday’s Systems of Record to what we have now:
“Enterprise software has moved through multiple generations, evolving from basic data storage tools into dynamic systems that drive collaboration and engagement,” Gottlob writes. “Initially, these systems served as Systems-of-Record, storing and organizing data. Over time, they progressed into Systems-of-Engagement, enabling users to interact with information and collaborate across teams. Today, we stand at the dawn of the next transformation: Systems-of-Work.”
Gottlob gives this detail on Systems of Work:
“AI agents, or ‘AI workers,’ handle tasks autonomously, engaging in workflows, making decisions, and producing outcomes. These AI-driven workflows eliminate repetitive tasks, freeing human teams for high-value, strategic activities. In this paradigm, software is no longer passive; it becomes a digital workforce that actively delivers results.”
So the AI is the default, as Frank Diana notes in writing about his concept at his blog, Reimagining the Future:
“Historically, businesses bought software and hired human professionals to operate it,” Diana writes.
“AI is flipping that model. Increasingly, businesses will design their processes around AI from the start – not as an add-on, but as the default worker.”
Applications to Business
It’s not hard to imagine how all of this will change business. Just think about the utility of Services as Software in consulting and managed services, and automating items like strategy audits, maturity assessments, and compliance checks with AI agents, or using the “copilots” for IT ops, marketing ops, RevOps, and HR onboarding.
However, like any new technology, this approach has to be done right, integrated into business as usual in a helpful way. Any big change can help or hinder a business, according to how it’s implemented.
Just for fun, here’s what ChatGPT gave me when I asked about the difference between yesterday’s SaaS, and today’s SaS. First, the model politely acknowledged that this stuff is confusing:
SaaS and “services as software” kind of rhyme, but they’re almost opposite motions
Then it provided this fun contrast:
SaaS: “We built software; we’ll run it for you and you rent it.”
Services as software: “We used to sell hours and humans; now the service itself is a productized system (often software or AI) that scales without more humans.”
That’s probably as good a comparison as any.
I liked this part: “we used to sell hours and humans…” That really says it all. Human labor was a commodity that has seen its zenith, and now….
Stay tuned.
