How SaaS will die
For 20 years I’ve been working with SaaS systems - most notably Salesforce. A few weeks ago I started seeing posts on LinkedIn making claims that SaaS will Die. At the time I dismissed it as click bait, but I think there might be something in it and I can see a future where the traditional SaaS model is obsolete. In this post we’ll explore how this might play out and what you can do to adapt to a new SaaS-free world.
Introduction
I was sat at my desk in a company called I was working for at the time called Celerant in their internal IT function. I was responsible for all of their internal IT enterprise architecture and we were looking at implementing a CRM system, which was very important for business. Something had been implemented previously, but it was a complete flipping disaster - a traditional piece of software that you put on a CD. We needed to rip it out and start again and we were evaluating two options: one was Salesforce and the other was Microsoft CRM. At the time Microsoft CRM time was not really much more than an extension on top of outlook.
I remember reading about Salesforce's “no software” thing and the fact that it could be accessed over the internet and then I remember seeing that they had this Soap API which allows you to programmatically access their system and be able to interact with their data and make updates to a create new records and delete stuff - all over the internet and that was kind of revolutionary at the time. You couldn't just do that with any other business system around that time. I mean it sounds crazy now, it's taking as a given, but for commercial software, it was unheard of. I started getting a bit excited by this.
I asked my team to go and have a look at Microsoft CRM and Salesforce and come back to me with what they thought and my chief architect at the time, sent me an email which I will never forget: It had two images with two words under each image. The first image was of an one of these old 1920s gramophone things, you know the ones with the big horn that comes out of it. Underneath it was the word Microsoft CRM.
The next image which had Salesforce written underneath, was off an iPod. Back in 2005 iPods were the things to have. It was the state of the art technical gadget that everybody had to have and that one image perfectly encapsulated everything that I was feeling about Salesforce at the time and I instantly knew - BAM! I'm going to focus my career on that, on Salesforce.
Microsoft CRM (2005)
Salesforce (2005)
Why SaaS worked
The world before SaaS
You got a CD with a version of the software on it—and that was it. That was your starting point. But of course, once you ship that CD, you're not just supporting that version. You're also supporting the last one. And the one before that. And probably one from three years ago because some massive customer still hasn’t upgraded.
And then you get the call from Tesco. Their system's down. They're furious. You ask, “What version are you on?” They tell you. You check. “Ah... right. That’s two versions out of date.” And now you’re the problem. Because telling Tesco, “Well, you should’ve upgraded,” doesn’t fly. Now they’ve got to start a three-month internal change process just to get the right CD into the building—never mind actually doing the upgrade.
That was the reality. You couldn’t just push an update and move on. Everything was slower and more painful.
But the real problem we had was integration.
Speaking as a software guy, integrating different systems together has always been one of the biggest pains you can imagine. Historically, it took an absurd amount of time just wiring things up in different ways. Back in the day, we had these Enterprise Service Bus setups. You didn’t have cloud-based software; everything was on-prem.
You’d get a CD, install the software on some server in your building, and hook it up to your LAN—your internal network. Then you might have another piece of software whose sole job was to somehow integrate the data from System A (from one CD) with System B (from another CD).
If you were lucky, you could do that at the application level. But more often than not, you were just talking to some SQL database directly—pure database-level integration.
And of course, all of that only worked because everything was on your LAN, in one place, in your own private data center.
What SaaS promised
Where Salesforce got it right was by implementing multi-tenant system properly - they kind of defined it for the business space. You have one single tech stack, and one code base deployed for all of your customers and the magic they did of course was to say “we have our one single code base but out customers can have customisations that we will enable through our own product, without having to have different code bases. It makes complete sense because of course if you do that then you can get everybody in your R&D team to just focus on this one code-base and make the whole thing much better for everybody.
When Salesforce came along, one of the things that really stuck out to me was the fact that you had the ability to fetch data and interact with it programmatically over the internet. Back in 2005, nobody was doing that. It was completely—completely—revolutionary for business systems.
And if you could do that, it opened up the possibility of having one SaaS system talk to another system. That might be your own internal infrastructure, or it might be some other SaaS product—Marketo, Workday, whoever was around at the time. You could potentially integrate all of those things together. Again, more at the application level and less so at the process level, but still.
I think that innovation obviously came from the consumer space, but the idea of having SOAP-based APIs, accessible over the internet, was the single biggest thing that enabled the proliferation of SaaS. And of course, everything else about SaaS made perfect sense from a business model point of view
The state of SaaS today
Integration is still hard today, by the way. I think that’s part of how we ended up with this model where there’s been a lot of consolidation in the SaaS market—especially over the last 10 or 15 years.
You’ve had companies like Salesforce going out and acquiring loads of smaller organisations, and then folding those products into the Salesforce ecosystem.
I’ve got a story for you about Salesforce.
So, I started working with Salesforce back in 2005, and I distinctly remember going to this Salesforce event. It was at the Landmark Hotel in London, and I remember it clearly because it was the same hotel where that poor Russian guy was poisoned—just a few weeks before, I think—and sadly died. Allegedly, it was the Russian secret police or whoever, there was a whole thing about it.
Anyway, we were in that same hotel. Marc Benioff, the founder of Salesforce, was on stage giving this talk. He was going on about how Salesforce was going to take on—and outcompete—the big incumbents at the time: Microsoft, SAP, Siebel, and whoever else was in the mix back then.
Their whole pitch was basically: look, these companies don’t stand a chance. All their products sit on top of 25–30 year-old infrastructure and software. We’re not like that—we’re new, we’ve got fresh ideas, modern technology, and we can move fast.
And honestly, that’s pretty much what happened. But fast forward 20 years and now Salesforce are the incumbents. They’ve become the Microsoft of this generation.
Their business has evolved, no doubt. But the innovation? I don’t think it’s really kept up. And now, I think they’re in a position where they’ve lost that edge. Regaining it would be so hard, it’d probably break their entire business model. I don’t think this issue is unique to Salesforce - but rather the SaaS sector as a whole, because the larger your business becomes the more difficult it is to quickly adapt to change.
The pressure that SaaS faces today
Automation tools like n8n and make.com are making it easier than ever before to connect sophisticated AI driven workflows with disparate systems. This makes it easier to integrate system together creating the possibility of using smaller more focussed point solutions specific to the needs of the business that can be integrated into business processes.
AI system are becoming much more capable of working with software and other AI systems. The development of tools and common standards such as MCP are making it increasingly more easy to connect AI systems with tools.
Waste & Complexity - Companies like Salesforce which 20 years ago was basically just Sales Force Automation (SFA), have added to their capabilities through both organic development and acquisition. The consolidation of the SaaS market has mean that there are a smaller number of very large SaaS vendors with offerings that extend way beyond their original rationale. I believe this creates waste, because customers use fewer of the features available to them because they’re not needed. Also, there are now a bewildering array of options for licences, feature activations, and legacy features that can’t be deprecated.
Consumption based business models - SaaS was built on user seat based subscription models - get more eyes using your application and the more revenue you will generate. Today however the world is moving towards a different pricing model based on consumption, driven by the AI vendors pricing their services based on the consumption of AI generated content. SaaS vendors, who need to offer AI services in order to remain relevant, will need to either align these two very different pricing models or just adopt the consumption based model instead.
What does a world without saas look like?
What does a world without monolithic SaaS platforms look like?
Imagine a landscape of smaller, niche service providers—each one focused on doing a very specific task exceptionally well. Instead of buying a “Quality Street” tin of features you’ll never use, you pick and mix the exact tools you need. As adoption of standards like MCP (Model Context Protocol) grows, interoperability becomes a given—and vendors will have no choice but to support it if they want to stay relevant.
Data lakes will still pose challenges—especially around time-based analysis—until we solve the “time travel” problem of capturing every state change. But with AI agents orchestrating everything, pulling data from multiple sources, enriching it on the fly, and synthesising new insights for each task, those silos start to dissolve.
That world may lie just around the corner. Smaller, best-of-breed services connected by intelligence-driven workflows could replace many of today’s sprawling SaaS suites—giving you the freedom to build exactly the stack your business needs.
What should you do about it?
Pilot “pick-and-mix” automations with n8n or Make.com
Identify one or two high-impact processes—such as cold outreach research or social media content generation—and build small, targeted workflows that connect your best-of-breed tools. These quick wins will demonstrate how niche services can deliver value when orchestrated by lightweight automation.Hold your SaaS vendors accountable on AI interoperability
As new standards like MCP (Model Context Protocol) and A2A (Agent to Agent) gain traction, demand concrete roadmaps from your platform partners. Ask how they’ll support standardised AI agent integrations and tool usage so you can continue to assemble the optimal tech stack, rather than being locked into monolithic suites.Reassess pricing models through a consumption-based lens
Traditional seat-based subscriptions are becoming less common and being replaced with consumption based models that better align to the economics of serving AI systems. Begin modelling total cost under usage-based pricing—whether API calls, data processing, or transaction volumes—so you can begin to understand the financial implications of adopting these usage based systems in the future.
