A while back I posted a short version of this thought on X: the original post. It got more of a reaction than I expected, so here's the long version — because the list of things I've replaced instead of subscribing to has only gotten longer since then.
What I've replaced so far
It started with flashcards. Every Anki app wanted $15-35/month for what's basically a CRUD app with spaced repetition. So I built my own — login, decks, cards, the browser's speech synthesis for pronunciation, an AI check for typed phrases. My wife and I have used it daily for months. It costs about $2/month to run.
Then it was budgeting. I wanted YNAB's method without YNAB's rigidity — its category rules, its onboarding, its opinions about how I should think about money. So I built my own version: same idea, none of the constraints I didn't ask for.
Then nutrition tracking. Same story — the paid apps in that category are either ad-supported, subscription-gated, or both. Mine isn't. A workout log is next; it's an obvious extension of the same app, and I already know exactly what I want from it because I'm the only user I have to design for.
AI does the reading for me now
I barely read the news anymore — not because I stopped caring, but because I set up AI monitoring for the handful of topics that actually matter to me, and it surfaces what's relevant instead of me scrolling for it. That's the same shift, really: a generic product (the news feed, the app store listing) replaced by something scoped exactly to one person.
I use AI in almost everything I build now, and it's not subtle — things that used to take a weekend take an evening. The gap between "I have an idea for a tool" and "I'm using the tool" has basically collapsed.
Where this goes for SaaS
Here's the part that isn't just about my own workflow. A lot of people are chasing the same playbook I half-am: build a SaaS, get it to some large monthly number, sell to everyone who has this problem. That playbook assumes the problem is hard to solve yourself. For a growing list of categories, it isn't anymore. If your product's entire value is "generic features, nicely packaged, behind a subscription," you're competing with something that costs one evening and a few dollars a month to build — and unlike your roadmap, it's built for exactly one person.
I don't think every SaaS dies. Anything that needs real infrastructure, network effects, or genuine expertise is fine. But the category of "app that wraps a simple idea in accounts, a paywall, and a pricing page" is exactly the category that gets eaten first — because that's the category an evening with an AI can now fully replace.
The reflex has changed
I notice it most in how I look at other people's apps now. I used to see something useful and think about buying it. Now I look at it and think: okay, these two or three features are the actual value — I'll build my own version, put it on my own server, hang it off a subdomain I already own. Not out of stubbornness. It's genuinely just faster than evaluating a subscription, a trial, and a cancellation flow.
And yes, I'm also selling one of these
Some of what I build stops being "just for me" — Screo is the clearest example, a CleanShot X alternative I built because I wanted the workflow without the subscription, ended up liking it enough to actually ship it. I use it every day. Maybe other people will too. The irony of writing "SaaS is in trouble" while trying to sell one isn't lost on me — good luck to me, I guess.
Same principles as everything else I ship, though: no accounts, no cloud, nothing stored anywhere but your own device. If I'm going to be part of the wave replacing subscription software, I'd rather not also be the thing people are trying to get away from.