Is the future software that bends to your company, not the other way around?
Deposit Photos
In February, SaaSpocalypse roiled markets, wiping out roughly $285 billion in SaaS valuations. Claude Cowork plugins were the culprit for the massacre.
To appreciate why, think about it like this. Until then, most companies required software programs to handle high-value knowledge work like legal analysis, sales operations, finance, and more. The Software as a Service (SaaS) industry built its revenue model on charging subscription fees to businesses for this valuable work.
However, thanks to Anthropic’s plugins, businesses could now create their own AI agents to spin up software cheaply. Not only that, but they could also customize such software, making it do what they want, how they want it.
But there’s more to the SaaSpocalypse saga. To appreciate what’s coming, it’s helpful to think of software eras.
Software Era 1: Packaged
It used to be you had to buy software on a disc or some other external device and install it on your computer where it resided on your hardware. You might keep that software version on your computer for several years before you had to buy the next release and then repeat the same process.
Software Era 2: Cloud
Under CEO Marc Benioff, Salesforce kicked off its “End of Software” campaign in 1999. Per The Drum, it “didn’t exactly end software, it did sweep in changes to many of the clunky, hard-to-manage tools that came before it.” Instead of installing software from an external source, the internet delivered it. Licensed by the seat, it could be updated continuously and digitally.
According to Philipp Wehn, we are entering the third era: specialized software that bends to your company instead of your company bending to it. The founder and CEO of Silicon Valley-based Nexxa.AI told me what this means in an interview.
Imagine a $30 billion industrial supplier learns of bad news: a shipment of material is running insanely late. This supplier cannot accept such tardiness. It has a promise to a carmaker that it must deliver in two weeks.
Before the third era, a frantic scramble would likely ensue.
The scheduler would have to use disparate programs to coordinate the crisis: One program to track the missing shipment, another to check in with the carmaker, one more to deal with the part design, and a fourth to determine what the factory floor could produce within the closing window of time. On top of all that, the scheduler would have to message at least three other coworkers, roping them in by spreadsheet, managing it all on endless Zoom calls.
In the solution Nexxa.AI is building, the scheduler turns to a single system that handles everything at once. As Wehn describes it, an AI layer reads across every one of the company’s programs at the same time, threads the information together, drafts a fix, and updates the records itself.
The seeds of such effectiveness in manufacturing are something I recently covered for Forbes via the lens of Agentic AI. It is reshaping workflows, providing an intelligence layer that allows companies to do more with less.
In Wehn’s vision, it’s not that a generic AI chatbot is bolted onto old software. Rather, AI is being used much like a brilliant new hire thrown into the mix, if that hire had a Mensa-level IQ and already knew what everything does in your company and could peer into every system at once.
To understand how this works, we must think in threes again. Nexxa operates in three phases.
Phase 1: Nexxa layers over the software a company already has in place.
Phase 2: As Nexxa’s AI does more of the work, users stop opening the underlying programs directly. The old screens and menus matter less and less.
Phase 3: Eventually, Nexxa’s AI learns enough about the underlying programs, such as their logic, connections, and the context they hold, to replicate them.
In other words, the old generic software retires, not because a newer disc arrives in the mail, and not because the cloud pushed an update. Instead, it occurs because the intelligence layer has learned to do the job itself.
The software has finally bent to your company, not the other way around.
Schaeffler Group, a German industrial giant that produces precision components, including the ball bearings and powertrain parts carmakers depend on, sees the value in the third era of customizable software. “Our partnership with Nexxa reflects Schaeffler’s commitment to shaping the future of industrial AI,” said its CEO Klaus Rosenfeld. “Together, we are combining decades of engineering excellence with cutting-edge intelligence to set new standards for how global manufacturers operate.”
Changing gears, no pun intended, from industrial applications to wider implications, we can look to another company that long ago recognized the need for changing how work gets done. Roughly 15 years back, Elon Musk decided off-the-shelf business software didn’t fit the bill for Tesla.
So, what did his company do? It created its own software, called Warp, as in Star Trek’s warp drive.
Chief Information Officer Jay Vijayan explained it in 2014 for CIO Insight this way: “Elon’s vision is to build a vertically integrated organization where information flow happens seamlessly across departments and where we have a closed feedback loop to our customers. By doing this, we can provide the best possible product, service, and overall experience to our customers in the fastest way possible; while also operating efficiently as a business to bring this vision to life, we had to have simple and central business operations software that could connect all departments and enable information flow seamlessly across departments.”
Since Tesla couldn’t rely on just any software program in the market to satisfy its need, Tesla built its own. But that was Tesla. It could afford to throw vast sums of money at the problem.
Now in 2026, if Wehn is correct, everyone will be able to do this, even small business owners who have struggled with stodgy software that comes nowhere close to delivering the value they need.
What this bodes for the future is not the AI doomer narrative so often on display in the media. A future in which every business can customize its operations through an intelligence layer might sound more like science fiction than fact, until you realize it is increasingly possible through AI.
Yes, SaaSpocalypse was incredibly disruptive to the market. Its deleterious effects are still being absorbed by software companies. But what may shake out when the dust settles is a more customizable way of working that allows more companies to deliver value more effectively. That cannot help but usher in more productivity, and ultimately, more prosperity.
