Is selling software a fake problem now? The thread says the fake problem is that question
Source: Will selling software become a fake proposition? (V2EX, ~5,461 views / 36 replies)
The OP quoted Meng Yan: once AI is strong enough, anyone can talk an app into existence for daily needs—no waiting for a big company’s one-size-fits-all product—so making money from apps gets harder. Then: will PMs, programmers, and designers slowly disappear?
Almost nobody in the 36 replies fully bought “selling software is dead.” The slogan that stuck:
People can cook. Restaurants still exist.
What the thread argued
Can ≠ will ≠ can do well
There are only 24 hours. Stability, support, and ongoing maintenance still sell. Wall-scraping, rice cookers, “you could hand-build a car but would you take it on the highway?”—same point: most people can muddle through; few want to, and fewer trust their own build.
“Don’t overrate ‘everyone’”—many won’t even open their mouths when food is already at the lips. Real vibe-coding is still niche.
The bottleneck moved: building is easy; being found is not
For ordinary apps the hard part hasn’t been code for a while—it’s distribution: ads, KOLs, channels. SaaS with private data doesn’t get swapped lightly. AI raises personal productivity; organized companies still ship stabler, nicer products.
Blunter take: AI only raised the bar for selling software. Half-day vibe tools lose the market; Chrome, ffmpeg, AAA games—you can’t afford the time or knowledge even if tokens are free.
The form already shifted: selling software ≈ selling time / service
B2B has long sold services. Some expect a new normal of “paste a ¥100 one-off,” but after you serve one client and distill the brief, the next job compounds—better than a fresh two-hour housecleaning every time. Hardware+software, licenses, and business models are harder to swallow than “can the model write code.”
Caveats: mini-program filings, Apple fees, tightening Android permissions block the “everyone deploys” fantasy; “sounds more like indie micro-apps lose the market”; someone seriously asked for a vibe-coded Pixel Cake clone.
Same ruler as “software isn’t valuable”
Earlier I wrote Is software worthless in the AI era?: the community agrees what depreciates is the feature layer everyone can vibe; moats move to domain depth, data, workflows, trust, hardware, and credentials. I pushed further—when Agents are strong enough, even customization may commoditize.
Today’s post is the seller-anxiety cut of the same thesis:
| This thread | Software isn’t valuable |
|---|---|
| Is selling software fake? | Can software still build an edge? |
| Restaurant / wall analogies | Kill-line layering |
| Hard part: distribution & service | Hard part: trust, data, responsibility |
| Tiny tools harder to sell | Shells and settings die first |
Read together: the fake proposition isn’t “selling software”—it’s “everyone can build → nobody buys.” What collapses is pricing power on commodity tools; what doesn’t vanish overnight is people who pay for time, stability, and someone to own the outcome.
One more link: Good tech ≠ making money. Even if vibe punches implementation cost to zero, tech is still one link on the chain where customers pay. Distribution, trust, and stating the problem clearly were never free gifts from a stronger model.
Unsappy close
- Tiny / lookalike apps: harder to charge for—just admit it.
- Service, maintenance, private data, channels: still reasons to sell.
- Roles don’t vanish—they squeeze toward “define the problem + distribute + take the blame.” Talking a runnable demo into existence is not talking a ten-year business into existence.
Meng Yan’s line spreads anxiety; the thread mostly corrects with common sense. Selling software isn’t dead. What’s dead is the fantasy that shipping equals someone paying—and that fantasy was never very alive before AI either.
Comments