As I’ve used AI more in my daily work, I see the existential problem for basically all AI and SaaS startups:

Claude and Codex are both so good, that I don’t need your startup at all. I can pretty much roll my own in a matter of hours. I can build my version and have no need to pay for yours. Your competition is my imagination.

Literally my imagination. Not my imagination and the bandwidth of an engineer or other folks to help me make it so. I can compete with you myself for my own personal needs.

This should be something everyone building a new product should be thinking deeply about. Your prospective user (and their ability to build what they need with AI) is an indirect competitor now.

If we don’t acknowledge this reality, a lot of startups will fail quickly, and a lot of startups will be funded abundantly with no business taking that money.

But I don’t think SaaS is dead. I don’t see a zero-sum game here. There is a reason UIs exist - because, in some cases for some people, they are a far more pleasant (and in some cases, trustworthy) way to solve a problem than by talking to a chatbot.

Instead, I think the idea of a SaaS unicorn is dead. There is no longer one clear winner in a space, but rather, thousands - even billions - of options.

There will be some exceptions, market winners who somehow achieve such status by ticking literally all of these boxes:

• strong incumbent brand power, or can disrupt existing brands with power
• exceedingly reliable at scale
• deep data moat
• best-in-market consumer-grade UX design
• deeply principled & opinionated, almost personal, in their approach

But alongside them, billions of SaaS products will exist in the smallest, weirdest niches because they’re so easy to spin up, and very few will singularly become unicorns. Many will only have a few dozen/hundred users and be run by single people.

And that can be a fun world! I spun up my own app a few weeks ago, having nothing at all to do with my work at @Buffer, in 2 days, and I have a few hundred people using it monthly!

This is an opportunity for an abundance ecosystem for software, not destruction.

“Winning a market” now requires intensely deep thought about what value one can offer to potential users, when your potential customers can roll their own in a matter of hours.

But that doesn’t mean your product idea has no right to exist. Might just need to adjust expectations a bit. Embrace your weird.

As an ending note, it feels very weird and unnerving to talk about AI in this way as a music artist who does not use AI in the creation of my art. I entirely rely on my own human expression for that. It’s challenging to work through this when those same AI capabilities are being used to commodify human expression (including in code - not just what we’ve traditionally called art forms). I haven’t figured out how to deal with this yet.