Is SaaS dead because of AI?
No. SaaS isn't dead, but seat-based software is under pressure as AI does more of the work itself and value shifts to finished work.
No. SaaS is not dead, but the model it was built on is under pressure. Software sold by the seat assumed a person would do the work inside it. As AI does more of the work itself, value moves from the software a person uses to the finished work a business pays for.
Why are people asking whether SaaS is dead?
For years, most business software was sold per user, per month. The model fit a world where software was a workplace: the more people a customer employed, the more seats it bought.
AI breaks that link. Once AI finishes tasks on its own, the software can do more while fewer people log in to it. A pricing model tied to the number of users starts to drift away from the value delivered. That gap is what sits behind the question.
Which parts of SaaS are under the most pressure?
Not all software is exposed in the same way:
- Narrow apps that help a person do one step faster. If an AI system can do the step, a screen that helps someone do it is worth less.
- Per-seat pricing where value no longer tracks users. Many vendors are moving to blends of seats with usage, capacity or results, because seats alone no longer describe what customers get.
- Less exposed: systems that hold critical records and workflows. Software that stores the data a business runs on, and enforces how work moves through it, becomes the ground AI agents work on. It changes shape rather than disappearing.
So the honest answer is narrower than the headline. A specific way of building and pricing software is being squeezed. Software itself is not going away.
If not seats, where does the value go?
To the work. For every dollar spent on software, six are spent on services. Software helped people do the work. It never did the work. AI can now run large parts of it: drafting, checking, reconciling, scheduling, answering, following up.
That opens a much larger market than software ever had: businesses that sell the finished work itself, a closed set of books, a resolved case or a processed claim, rather than access to an application. We call these AI-native services companies. AI executes the routine. Specialists own the judgment. The customer buys the result.
What is the difference between selling software and selling the work?
When you sell access to software, every improvement in AI models makes it easier for someone else, including the customer, to build something similar. The advantage of having built the product shrinks over time.
When you sell the finished work, the same improvements make the work cheaper to deliver, and that gain stays with the business doing the work. It grows further when every specialist judgment is captured and built back in, so each job lowers the cost of the next. We explain that loop in How does a data flywheel make each job cheaper in AI-native services?
Pricing decides who keeps the gain. A business that is priced by the result keeps it. A business that charges by the hour or the seat hands it to the customer.
What should buyers and investors watch for?
For buyers, the useful question is no longer which software to license. It is whether you are paying for access or for the result you need, and whether the price follows that result.
For investors, the signals sit below the label:
- Does the business sell the work or access to it? A company that delivers finished work owns the gain when AI gets better.
- Is AI built into the work or added on top? Adding AI to the old workflow makes costs fall once. Rebuilding the work makes them keep falling.
- How fast does the work get cheaper? The measure is the human time behind the same job, tracked over time.
Questions people ask
Will AI agents replace SaaS?
Not wholesale. Agents need records, data and workflows to work in, and much of that lives in software. What changes is who does the work inside the software and how it is priced.
Is per-seat pricing going away?
Not everywhere. Seat pricing still fits software that people use directly every day. Where AI does the work itself, pricing is moving toward usage, capacity or results, often blended with seats.
What comes after SaaS?
There is no single replacement. A growing share of value goes to businesses that deliver finished work, with AI running the routine and specialists owning the judgment, priced by the result.
Is SaaS dead for small businesses?
No. Small businesses still rely on software to run their operations. The bigger change for them is that they can increasingly buy finished work, such as resolved support cases or kept books, instead of software they have to operate themselves.
Questions or ideas? Email hello@skywakelabs.com.
Buyers, investors and partners: hello@skywakelabs.com
More notes
What is agentic BPO?
Outsourcing where AI agents work most cases and specialists take the exceptions. Real agentic BPO prices the finished case, not seats or agents.
ReadWill AI replace procurement teams?
Not the team. AI takes the paperwork of buying; people keep the decisions with no written answer: which supplier, on what terms and whether to buy at all.
Read