人工智能并没有扼杀SaaS，而是正在淘汰单一功能的SaaS。
在过去的一年里，我不断看到“软件即服务（SaaS）已经死了”的观点。我并不认为情况是这样的。
目前受到压力的并不是SaaS作为一种模式，而是那些围绕单一功能构建的狭义SaaS，这些功能现在可以被人工智能以低成本复制。
如果你的产品基本上只是对一个模型的简单包装，或者主要通过功能而非工作流程集成来区分，那么现在的护城河感觉就会变得更弱。人工智能显著压缩了构建时间。这意味着竞争者增多、克隆速度加快，以及更低的转换成本。
然而，那些与模型无关、深度嵌入工作流程的SaaS，或者作为系统之间的连接纽带的SaaS，看起来要更具韧性。集成、分发和信任的贬值速度远不及功能的贬值速度。
这更像是SaaS的排序事件，而不是崩溃。简单的包装被挤压，而基础设施层和集成商则变得更强大。
我很好奇现在正在建设的其他人是否也看到了同样的变化，还是我对人工智能原生工作流程的关注过于偏重。
查看原文
Over the last year I keep seeing “SaaS is dead” takes. I don’t think that’s what’s happening.<p>What seems under pressure isn’t SaaS as a model. It’s narrow SaaS built around a single capability that AI can now reproduce cheaply.<p>If your product is basically a thin wrapper over a model, or differentiated mainly by features rather than workflow integration, the moat feels weaker now. AI compresses build time dramatically. That means more competitors, faster cloning, and lower switching costs.<p>But SaaS that is model-agnostic, deeply embedded into workflows, or acts as connective tissue between systems looks much more durable. Integration, distribution, and trust don’t commoditize as quickly as features do.<p>It feels less like SaaS collapsing and more like a sorting event. Thin wrappers get squeezed. Infrastructure layers and integrators get stronger.<p>Curious if others building right now are seeing the same shift, or if I’m over-indexing on AI-native workflows.