Ask HN: Any example of successful vibe-coded product?

Hacker News Discussion ↗

Range of vibe‑coded products mentioned

  • Numerous small to mid‑sized projects: browser and Windows extensions, video thumbnail generators, chat clients, word/puzzle games, Pomodoro tools, spreadsheet/Tabata/exercise apps, price‑tracking tools, polling visualizers, grocery price dashboards, and home‑automation controllers.
  • Several niche or local‑scale SaaS / web apps: CRM systems, restaurant POS, bar and lab inventory tools, educational platforms, data‑entry/reporting tools, AI config generators, chart‑from‑screenshot service, kids’ activity aggregator, recipe and media apps, and various marketplaces.
  • Some have small but real user bases (hundreds of users or daily players); a subset report paying customers and recurring revenue, but none claim large‑scale breakout success.

Personal vs commercial “success”

  • Many builders define success as “it works in production for me/my team” or “it replaces a subscription,” not VC‑style growth.
  • Several tools have been used continuously for months to a year or more and are considered stable and productivity‑enhancing.
  • One thread criticizes the lack of obviously big, commercially successful vibe‑coded products given the hype.
  • Others counter that a lot of value is in invisible internal tools and SaaS replacement, where success is measured in saved contracts and integration time, not public traction.

Definitions: vibe coding vs AI‑assisted coding

  • No consensus on “vibe coding”:
    • Strict view: never write or read code; the model does everything.
    • Looser view: the model writes most code; humans provide specs, review diffs, and fix issues.
  • Some insist AI‑assisted programming is distinct from full “vibe coding” and that most serious work is in the former category.

Perceived strengths and use cases

  • Best fit reported for:
    • Prototyping and MVPs.
    • Small utilities and tools in familiar stacks.
    • Internal CRMs/CMSs and workflow apps tailored to a specific mental model or business.
  • Experienced developers report large productivity gains when combining domain knowledge with LLMs, especially for greenfield work and front‑end/UI they’d otherwise avoid.

Limitations and skepticism

  • Several note that fully vibe‑coded apps tend to have rough UX and reliability issues; complex or large proprietary codebases remain hard for LLMs.
  • Common pattern: LLMs generate most scaffolding and boilerplate, but humans still debug logic, refine algorithms, and enforce tests.
  • Some argue LLMs are a multiplier for competent engineers, not a replacement; hiring practices at major AI companies are cited as evidence that deep engineering skill is still required.