Look up "SaaS is dead" and you'll get a list of posts and videos mostly agreeing with the claim. This has been going on since at least late 2024, when Satya Nadella's comments on a podcast were (mis)interpreted into the headline "SaaS is dead."
Almost two years later, SaaS is still very much alive, and so is the trend to read its eulogy.
What's worse is that this idea keeps getting conflated with the idea that software engineering as a profession will collapse. It's not just that SaaS is dead. It's that nobody needs software engineers anymore, because AI will build their tools for them.
Only time will tell whether either headline proves true. Until then, I, for one, don't think SaaS is anywhere near dead. Nor do I think that software engineering is dead. In fact, I think software engineers are poised to thrive.
What's on its way out is charging money for trivial solutions.
Weekend builds vs. engineered products
AI has made it remarkably easy to build something that works. A weekend and a few prompts, and you have a tool that solves some specific problem.
However, "works for me on a Sunday" is a long way from a product. Engineered products carry things a weekend build doesn't: reliability, security, integrations, and maintainability.
Agents can't easily replace those. What they can replace is a subscription to an app that does one small thing.
A higher bar is good for everyone
The bar for charging money is rising. Whether a product can build a healthy pool of users, and with it a sustainable business, increasingly depends on three factors:
- the difficulty of the problem you're solving
- the difficulty of maintaining the solution
- the demand in the market
Demand is a prerequisite. If there's no market, there's no business. However, for our purposes, the first two factors are the interesting ones.
In the age of agents, products that earn money are not ones that require many lines of code. Code is cheap now. They are the ones that are hard to build without engineering skills, or hard to maintain beyond version 1. You hit the jackpot when your product scores high on both factors.
A micro-SaaS that can easily be replaced by some n8n workflow, or a skill on your favorite AI assistant, will no longer cut it. The niche problem is not the issue here; the simplicity of the solution is.
Similarly, a slightly more demanding product that needs only to be built once and gets forgotten about, or gets used occasionally, will attract more people attempting to build it.
While this raises the barrier to entry, it also means that the market will force higher value. Higher competition leads to better products, lower prices, or both.
The sieve
"Solve your own problem first, then sell it" has always been a proven route to a product. AI does more than run it faster. It also tightens the sieve.
An itch that was easy to scratch used to be a product, because only engineers could build the fix. Now almost anyone can. So:
- If your weekend build solves the problem completely, it will be hard to sell.
- If you keep hitting walls, like edge cases, scale, data, or reliability, you might be onto something people would actually pay for.
Hitting walls doesn't guarantee anyone will pay. You still need other people who have the same problem. But the walls are a useful signal. Each one is a reason someone might rather pay you than build it themselves.
Code is not the product in SaaS
The last letter in SaaS stands for "service," and that's the part that gets lost in the "SaaS is dead" conversation. Customers don't pay for your code. Most of them will never see it. They pay for what happens around it:
- the app is up when they need it,
- it complies with standards and regulations,
- it stays fast as their data grows,
- their data is safe, backed up, and not leaking,
- updates ship without breaking their workflow,
- and someone handles it when something goes wrong.
Agents have made code cheap, but code was never the product. That's why "I can rebuild that in a weekend" can be both true and missing the point.
Why engineers come out ahead
I'm not a leader in the field. I'm just an average career switcher who loves programming. So I claim no authority when I say: I truly believe that engineers come out ahead in all this.
Software engineers are the ones with the skills to give agents real system requirements. They're the ones who can discuss architecture with an agent and, when they're not being lazy, review its output instead of letting an unmaintainable monstrosity grow from layers of agentic patches.
Software engineers are also the most able to scratch their own itches. So they get to practice what most of them love to do — building solutions — at an unprecedented rate.
And when a problem turns out to be hard enough, they're the ones best equipped to evolve a toy project into a viable product.
Easy itch or hard problem, engineers win either way.
Top comments (0)