"SaaS is dead" is a good headline and a poor description. Companies still pay for software and will keep doing so. What's changing is what they're willing to pay for, and that affects anyone building or buying software, or selling AI services.
What's actually changing
For twenty years, the model was simple. A company built a tool and charged per user per month. It worked because building software was slow and expensive, and because every extra employee meant another seat.
AI is weakening both of those assumptions.
Building got cheaper. One person using AI can now build a working version of a simple tool in days. When a basic tool is easy to copy, the tool itself stops being rare.
Fewer people need to log in. An AI agent can do the work directly, without using the screens the software company charges for. If ten staff become two staff plus an agent, a per-seat bill falls by 80% even if the work grows.
What the numbers say
In July 2026, Gartner estimated that up to $234 billion of enterprise application spending is exposed to agents between now and 2030, roughly 20% of enterprise application SaaS spending by then. Gartner's argument is that agents bypass user interfaces, which breaks the link between user growth and revenue. It also says vendors need to move from interface-based value to outcome-based value.
Read that number carefully. Twenty percent exposed means that most spending isn't, at least by that estimate. Serious software with deep data, security and compliance isn't going anywhere soon.
From selling tools to selling results
Nobody wants a subscription to a hammer. People want the house built.
The direction is toward three layers:
- The tool becomes cheap or free, and that's how people find you.
- The agent does the work on top of it.
- The service sets it up, runs it and takes responsibility for the result.
A business that promises "leads answered and appointments booked" is in a stronger position than one that promises "access to our platform", because the first one is what the customer wanted all along.
Don't overcorrect: agents aren't free
There's an opposing force. Gartner also predicts that the cost per resolution for generative AI in customer service will exceed offshore human agents by 2030. So "AI makes everything cheaper forever" is too simple.
Our own numbers on voice agents are low today, about $0.26 per answered call, but usage costs are real and worth watching. Anyone pricing on outcomes needs to know their cost per result.
What this means for you
If you buy software: check what you pay per seat. Ask which of those tools an agent could make unnecessary, and which ones hold data you can't afford to lose.
If you build software: a thin tool is a weak business now. Own something hard to copy, like your data, your distribution or a specific workflow you understand better than anyone.
If you're not technical: this is good news. The valuable skill has shifted from writing code toward knowing a business problem, telling AI what you want and checking the result.
What to try this week
- Pick one painful, repeated job. Answering the same questions, chasing leads, booking appointments. If it happens every week, it qualifies.
- Try the free version first. Judge an AI tool on one thing: did it save real time this week?
- If you sell to clients, sell the result. Don't sell a subscription. Sell leads answered within minutes, or hours saved.
The Dubai and GCC angle
Real estate teams, clinics and service businesses here don't want another login. They want enquiries answered, in the right language and at any hour. That's why we build AI lead systems and call assistants around a single measure: bookings.
If you want to make this your business, start with how to start an AI automation agency.
This article began as our response to Greg Isenberg's video on the same question, on his YouTube channel. The analysis here is our own.
Frequently asked questions
Is SaaS really dead?
No. It's changing. Simple tools sold per seat are under the most pressure, while software with deep data, security or compliance value remains strong.
Why are AI agents a threat to SaaS?
Per-seat pricing depends on people logging in. Agents can do the work without using the interface, so a company may need fewer seats even as its work grows.
What is outcome-based pricing?
You charge for a result rather than access, such as a resolved support ticket or a booked appointment. It needs a clear definition of the result, so both sides agree on what counts.
What is an AI agent, in simple words?
It's software that completes a job rather than just answering a question. You give it a goal, like answering new leads and booking them in, and it follows the steps using your tools.
I'm not technical. Can I still benefit from this shift?
Yes. You can use ready-made AI tools or learn to set up AI systems for businesses. Our free masterclass builds them live on screen.