"Be Different" doesn't work for building products anymore

Original Article ↗ Hacker News Discussion ↗

AI “slop”, noise, and competition

  • Some argue: if “AI slop” wins on uptime, accuracy, etc., it’s not actually slop; the market is just saying it’s good enough.
  • Others say the problem isn’t that slop wins, but that volume of low-effort AI products drowns out higher‑quality work.
  • A suggested tactic: ship something AI-accelerated and imperfect to survive/runway, then refactor into a better product later.

Building vs distributing software

  • Multiple commenters push back on the idea of a “Cambrian explosion” of meaningful new apps; app stores were already saturated long before LLMs.
  • The real bottlenecks are distribution, algorithms, reputation, and network effects, not the ability to write code.
  • Even “perfect” clones of Office/Jira/etc. would struggle because users are locked in, conservative, and follow incumbents.

Limits of vibe-coded / AI-built apps

  • Strong consensus that AI excels at CRUD, prototypes, integrations, and boilerplate, but hits a “complexity cliff” for:
    • Real‑time collaboration, editors, simulations, low‑level systems, advanced geometry/algorithms, regulated domains.
  • AI often chooses poor abstractions (e.g., giant if‑trees instead of rules engines), or simulations instead of math, unless guided by an expert.
  • Thus, AI helps experienced devs go faster but cannot yet autonomously build or maintain complex, novel systems.

Moats, incumbents, and niches

  • Many see technical differentiation in simple products as non‑moaty; AI just accelerates copying and drives profits toward zero.
  • Moats remain in: deep integrations, infrastructure/reliability, security, regulatory know‑how, and long‑standing trust.
  • One interpretation of the article: you must now target obscure/complex niches where AI cloning and casual competitors can’t easily follow.

Big companies, startups, and funding

  • Skepticism that AI makes big companies “move fast”: bureaucracy, risk aversion, and coordination remain the limiters.
  • AI appears much more transformative for small teams and solo builders than for large enterprises.
  • Some hope this pushes VC attention away from generic SaaS toward harder areas like hardware and biotech.

Quality, trust, and user behavior

  • In a world of many polished but unreliable tools, support, customization, and trust become key differentiators, especially in B2B.
  • Analogy: vibe-coded apps are like cheap AliExpress products—fine for low‑stakes purchases, avoided for important ones.
  • Several note that user acquisition, not coding, is often the true hard problem, and that this predates AI.