There’s a headline going around that SaaS is dead. Microsoft’s CEO has been the loudest version of it, arguing that most business apps are just a thin screen on top of a database, and that once AI agents can hold the logic themselves, the apps underneath fall away. It’s a clean story. For a second it even sounds right.
Then you try to actually do the work.
Here’s the sleight of hand in “SaaS is dead.” It swaps a demo for a product. Yes, a model can do something impressive once, in a chat, when someone technical is steering it. That is not the same thing as software that works every time, for a person who has never written a prompt, on their real files, on the hundred and twentieth try and not just the demo. The impressive moment and the dependable tool are two different things, and the distance between them is exactly where software lives.
Take something we do every day. You have a procedure to document, and you have a folder of screenshots. Not three. A hundred and twenty. Drop them into a chat window and it falls over. That is not what a chat window is for. It will truncate, lose the thread, or just shut you down. Now take those same hundred and twenty images, drop them into Notavi, and get back a clean, ordered, illustrated guide, every step in place, built by someone who has never thought about a prompt in their life. The whole thing takes a couple of minutes.
The AI is not the hard part of that. The hard part is everything around it. Capturing the steps as the person works. Keeping a hundred and twenty images in order and matched to the right step. Handling the one that failed to save without losing the others. Versioning, so you know which guide is current six months from now. Exporting something that does not fall apart in Word. Keeping it all on storage the user chose, so the work stays theirs. None of that is a thin screen over a database. It is the actual job, and a general-purpose agent does not do it for you just because it can write a sentence.
So here is the part the “SaaS is dead” crowd has backwards. AI did not kill that work. It raised the ceiling on it. A capability on its own is raw material. Turning a capability into something a regular person can rely on, every time, without thinking about it, is the entire discipline of good software. The thin wrappers, the apps that really were just a form on top of a table, those will go, and good riddance. The tools that do real work are about to get a lot sharper.
And there’s an irony worth sitting with. The same company telling us business apps are finished is spending billions building more of them, with AI inside, and asking you to pay for it. That is not the death of software. That is a very large bet that the software around the model is where the value is. We happen to agree. We’re just making that bet one procedure at a time, for the person who has always been the one everyone asks, “how do I do this again?”
SaaS is not dead. It’s being asked, finally, to justify itself. The ones that were always just a screen on a database should be nervous. The ones that do real work, that take a hard, human task and make it genuinely easier, have never had more room to run.
That’s the bet we’re making with Notavi.