AI 不会杀死 SaaS：四个理由与未来变化

All right, so NR's my colleague, he couldn't be here today cuz he's unfortunately sick, so I've taken his presentation from him um and all of his notes are the same. We will not be doing any questions though because I will all those need to go directly to him. But, him and I collaborated and uh wanted to bring you this quick little presentation on AI, which as we all know is a um pretty big topic right now. Whoop. There we go.

All right, so you've probably wondered right now in this AI world, will AI kill SaaS? Now, this is a serious question because we've been watching these capabilities explode, and depending on who you ask, you're either watching SaaS enter its golden years or its final act. And if especially if you're on Twitter Twitter doom scrolling on Twitter, people are really really uh you know, pessimis- pessimistic about the future of SaaS.

Because AI has been making code more cheaper more cheaper and more accessible, more people than ever can stitch together internal tools that meet their exact needs, and that's this has led to people canceling SaaS subscriptions. And we can make the assumption that in the next year or two, we can see AI automating entire categories of SaaS. So, if you think about it, why would you pay for a dashboard or a marketing optimizer when AI can directly handle tasks like optimize my ad budget or analyze Q2 results on demand.

And then we can make the assumption that in 3 to 5 years, software could could become invisible. AI is not going to just help you use tools, it will become the tool. It can solve end-to-end business problems with no app or SaaS platform in sight. But, there's a compelling counter argument, the case for resurrection. So, I'll go through these four key reasons why SaaS might not only survive, but grow more powerful in this AI-powered world.

So, first, AI capability improvements are not accelerating exponentially forever. The recent breakthroughs like ChatGPT were less about new algorithms, but more about brute-forcing scale. With more data, we are there is more ability for compute. So, here's the catch. There that we're Here's the catch. We're running out of internet text to train on. So, if you look at something like the neural scaling law, this is where you really see AI coming through.

It shows that every time we double the resources, we get progressively smaller performance gains. So, infinite AI growth is not going to happen. Second, just because AI is democratizing democratizing the means of production by making it easier to create, it doesn't mean that it automates creativity or excellence. AI can't theoretically, you know, write trillions of books, but it wouldn't know which ones are masterpieces.

True creativity often comes from breaking patterns, from knowing when and where to break the rules, and that's something that AI simply is not great at. Third, software complexity grows alongside societal complexity. Human desires and economic activity don't have an upper limit. They're infinite. And software has evolved to meet that growing complexity. So, when development gets cheaper, thanks to AI, we don't end up with fewer tools, we end up with more niche solutions and more tailored products.

So, this is is something that's called Jevons paradox, probably applies. So, that shows that as a resource becomes more efficient, the overall consumption of that resource will actually increase. So, we can make the assumption that as software becomes cheaper to make, we will build even more of it. And finally, this might be comforting, we're not creating God. Real-world problems don't have neat computable answers like chess.

They involve nuance, judgment, intuition, in infinite branching infinite branching plat- paths. Even the best AI struggles here, especially when interfacing with with like messy human consumption cognition. So, unless we're building an all-seeing, all-knowing AI that can read minds and predict the future, SaaS isn't going anywhere. So, where does this leave us? We believe that AI will quickly disrupt SaaS with simple, well-defined tasks.

So, if you're building something with simple, well-defined tasks, those are going to get wiped out really fast. But things that involve human creativity, empathy, and understanding, those will become even more valuable. These are the things that AI cannot replicate. The incumbents, they're more vulnerable than ever. And that's good news for everybody in this room. We're standing at this edge of a massive opportunity.

We're going to see teams of like five to 10 people building SaaS products that will be making hundreds of millions of revenue. And that's not just the distant future, that is coming soon, within the next few years. So, we're in this really exciting period of time. AI is not going to kill SaaS. There is a lot of really awesome things that we are right on the brink of. And that's it.

[applause]