I recently watched Wu Minghui's long interview. Fascinating.
Frankly, I've always been skeptical of grand narratives like "Agents are killing SaaS." The AI world has no shortage of tech evangelists and futurist preachers.
But there's something rare about Wu Minghui: you can feel that he actually believes it.
And not the PowerPoint-founder kind of belief. This is someone who has already taken a massive fall — his company nearly died, he laid off brothers, got brutally beaten up by reality — and yet somehow still dares to believe in the future again.
You can't help but have a soft spot for people like that.
What I found most valuable in his interview isn't the slogan "Agents are killing SaaS." It's three deeper points.
First: the software shell is rapidly depreciating.
When the requirements are clear, the interaction paradigm is mature, and the data structure isn't complex, an Agent + coding model can replicate traditional SaaS faster and faster. The software shell — built over years with engineering man-months, organizational discipline, and long cycles — is commoditizing at speed. For many SaaS companies, the biggest moat was never intelligence. It was implementation. And now implementation itself is being swallowed by models.
Second: real value is shifting from software to context, workflow, specialized models, and taste.
Going forward, what's valuable isn't "we built another Feishu/CRM/BI system." It's who owns the industry data, who understands real workflows, who can embed Agents into organizational collaboration, and who can build attributable, governable, sustainably iterative human-machine networks. Software is becoming the plastic casing. The context flowing through it is the real asset.
Third — and this is the most interesting one: Wu Minghui says "I think, therefore I am" is becoming "I taste, therefore I am."
Thinking is deterministic reasoning. Taste is direction, aesthetics, life experience, accumulated context. AI is rapidly devouring the former, but it's nowhere near the latter.
Many people aren't being replaced by AI in their thinking. They just never got around to forming their own taste. The truly brutal future may not be "AI takes your job" — it's masses of people discovering for the first time that decades of their work was essentially process execution, not judgment.
One more part that got me: he said that even if investors and the board push him to lay people off, he'll resist as much as he can — because if every company only optimizes for cost reduction, the demand side will eventually collapse.
Emotionally, it's moving. Logically, it's not entirely baseless. But the biggest soft spot is: without validating the Agentic Service business loop first, "no layoffs" is essentially a beautiful post-dated check. Supply-side technological leaps don't automatically create demand.
If Minglue really succeeds in not laying people off — or even hiring more — thanks to AI, it probably means they ate someone else's share. At a macro level, the vision of "everyone happier because of AI" feels a bit naive.
But here's the interesting part: I don't actually hate this naivety.
In an era of mass anxiety, where everyone fears being replaced, seeing someone who has experienced catastrophic failure still willing to believe so sincerely that "people still have value" — that alone is precious. The tech world isn't always pushed forward by the most coldly rational people. Sometimes it's pushed by those who know they might lose, but choose to believe in something anyway.
