Nearly $2 trillion in software companies’ market cap was wiped out in early 2026, and a lot of people are calling it a technology crisis. I think it is a business model crisis.
Investors have long priced SaaS companies at a premium because subscription revenue was predictable. Recurring contracts meant low downside risk, and the model itself was the moat. Now, capital markets are saying AI poses a fundamental risk to that model, and they are repricing accordingly.
Every software company is facing the same question right now: Are you being disrupted by AI, or are you using it to build something amazing that customers will love? The answer to this depends on whether you’re building only for humans or the hybrid human-agent workforce we are already seeing.
Need more clues? Ask the Sherlock chatbot (lower right corner) to summarize this story, explain technical concepts or answer other questions.
Agents do not work the way humans do
The entire SaaS model was built on the assumption that a human being is the primary user. A person enters data, validates it, manipulates it, and acts on it, all inside a single closed application designed around that workflow. The interface is the product, and the interface justifies the price.
Agents do not need an interface, dropdown menus, guided workflows, or dashboards. They have an objective and can figure out how to get it done by pulling from wherever the relevant data lives. When agents start handling a significant share of the work humans used to do inside these applications, the UX layer that justified the price tag becomes irrelevant.
Nvidia CEO Jensen Huang has said agents will use software tools, not replace them, and I agree. The tools and the data still matter. The difference is that agents interact with software completely differently than humans do. Traditional automation stays within a single system and agents reason across systems. That cross-system reasoning is where the category is being won and lost.
The shift from human-only to hybrid workflows doesn’t just happen overnight and it does not mean ripping out everything you’ve already built.
Practically, there are a few things worth thinking through.
First, ask yourself whether your platform is accessible to agents or whether it requires a human to operate it. Most enterprise software today assumes a person is clicking through it. If that assumption is baked into your architecture, then agents will route around you.
Second, ask where your proprietary data lives and whether agents can get to it. Public LLMs are trained on public data. The expertise that makes a business competitive lives in private systems, in the communications, deal histories, and relationship patterns that define how that business actually operates. If you try to throw generic AI at a complex enterprise function without that proprietary context, you get outputs like “build trust with a stakeholder” or “accelerate this deal.” Those answers sound reasonable and do absolutely nothing.
What the transition looks like in practice
The shift from human-only to hybrid workflows doesn’t just happen overnight and it does not mean ripping out everything you’ve already built. Instead, it means asking different questions about what your software really does and who (or what) needs to use it.
Most enterprise software companies are still charging for seats because that’s the unit that they know how to price and long have priced. But, when an agent handles the pipeline analysis that previously necessitated three analysts or automates the data entry that had in the past justified five CRM licenses, charging for seats that no longer exist stops making much sense.
The instinct is to lock data behind your own walls because that is how you have always controlled the customer relationship. In an agentic world, however, that instinct works against you.
The challenging thing is that pricing for value delivered is legitimately harder than pricing for user access. What value means must be redefined and you must be able to measure it, or at least that’s what the successful companies will do.
The companies in a strong position after this period of change will be the ones that make their proprietary data accessible to agents, which requires a different kind of openness than most enterprise software companies have historically resisted. The instinct is to lock data behind your own walls because that is how you have always controlled the customer relationship. In an agentic world, however, that instinct works against you.
Building for the new, hybrid workforce
Everyone assumed the web would eliminate whole categories of work. The activities did not go away. They changed in how they were done. The same transition is happening now.
The unit of value in enterprise software used to be access to a tool, but that shifts to the quality of the answer in the agentic era. Seat-based pricing defined SaaS for two decades. When an agent is doing the work that three people used to do inside your platform, it’s no longer tenable to charge for seats.
Software CEOs need to answer a harder question: What are we actually selling? The companies that figure out how to price for value delivered, rather than users logged in, will define what enterprise software looks like over the next decade. The ones that don’t will be remembered as casualties of a model that stopped making sense long ago.
The repricing happening in public markets right now is both rational and necessary. It is the market asking a question that a lot of software companies are not yet prepared to answer.
Be First to Comment