Vibe Coded Apps: 82 Cases With Revenue Evidence

Vibe coded apps are commercially credible when they solve a narrow paid problem and publish evidence, not merely when AI generated the code. ProvenStartups tracks 82 projects in this cohort: 59 are solo-run, and 22 publish a clean monthly figure with a $20K/mo median across the full matching set. Start with verified cases such as Payout at $20K/mo [V], and reject forecasts or undisclosed sales as proof.

Contents

This page moves from the cohort-level answer to named cases, then turns the evidence into a build filter. The distinction is simple: a shipped vibe coding app is a technical artifact; a viable business has a paid problem, a distribution path, and a revenue claim strong enough to grade.

• ·What the revenue evidence shows
• ·Vibe coded apps compared
• ·Where the data contradicts popular advice
• ·What we would build
• ·What we would refuse to build
• ·FAQ

What the revenue evidence shows

The cohort says vibe coding can produce real businesses, but disclosure must be separated from storytelling. Among 82 matching projects, 22 publish a clean monthly figure; their median is $20K/mo and their range is $2K/mo to $500K/mo across the full matching set, not only the named samples below.

The cohort includes 59 solo-run projects. Its stated evidence split is 16 [V], 0 [F], 0 [C], and 0 [U]. Those counts do not turn every remaining project into a verified claim; they show why the evidence label must stay beside each disclosed result.

Across the full index of startup ideas, ProvenStartups tracks 406 cases: 57 third-party verified [V], 184 founder-reported [F], 121 creator-relayed [C], and 44 unverified [U]. The grading method separates source quality from how attractive a business sounds. It also preserves 38 cautionary tales instead of quietly deleting the failures.

The phrase itself describes software creation led substantially by natural-language prompting. Wikipedia's entry on vibe coding gives the general definition, while Anthropic's Claude Code documentation shows one agentic coding workflow. Neither tells you whether users paid.

That distinction prevents a scale reference such as Cursor, at $500M/yr [V], from being blended with an early-stage claim. A big result is useful context, but it is not an automatic blueprint for a solo founder.

Vibe coded apps compared

Revenue should be read with source quality and business shape in the same row. A high number with a weak class is a lead to investigate, not a fact to repeat; a smaller [V] result can be the better model. These cases show the spread in outcome, complexity, and disclosure.

Payout and Subscribr are stronger models than a portfolio with undisclosed revenue because both pair a narrow job with a verified result. AEO Service is weaker evidence, but its $2K/mo [F] single-client retainer still establishes an actual transaction rather than projected demand.

Difficulty is not a revenue ranking. The table contains a 1/5 service, 3/5 apps, a 4/5 distribution-led SaaS operation, and a 5/5 infrastructure company. Choose complexity after proving the buyer, not as a substitute for proving one.

Where the data contradicts popular advice

Our data contradicts the niche's loudest claim: shipping more apps faster is not, by itself, a business model. Verified winners attach to painful jobs or existing demand, while the portfolio pitch often supplies launch counts instead of revenue. Code generation compresses production; it does not manufacture trust, traffic, or willingness to pay.

The AI Directory Site targets $2K-$10K/mo [C] by citing benchmark sites at $1M-$5M/yr [C]. Those are targets and borrowed benchmarks, not collected MRR. The AI Venture Studio describes profitability as theoretically uncapped [C], which is a thesis rather than evidence.

Revenue alone is not enough, either. NoFap reached $6K/mo [V] in its first month yet is filed as a cautionary tale. Fluently reported about $100 MRR [V] while barely breaking even on ad spend. A launch, revenue screenshot, and durable business are three different milestones.

The useful contradiction is blunt: one well-distributed workflow can beat a factory of thin wrappers. ProvenStartups would copy the validated pain and acquisition channel, not the number of repositories shipped.

What we would build

We would build one narrow workflow with an identifiable buyer, sell it manually, then automate only the repeated steps. This ordering fits the evidence better than generating a portfolio and waiting for discovery. For a solo vibe coder, distribution is the first system to design; the app is the second.

1. 1.Pick a problem with an existing budget. The AEO Service closed a $2K/mo [F] retainer from one client. That does not prove a broad market, but it proves one buyer paid.
2. 2.Sell the output before polishing the interface. AI SEO services for local businesses reported $5,000+ cumulative [F] from digital products, plus client retainers of several thousand dollars a month [F]. Service work exposes the inputs worth automating.
3. 3.Narrow the recurring job. Subscribr reached $30K/mo [V] around one repeated deliverable: YouTube scripts. Its boundary is clearer than a general-purpose “AI content” app.
4. 4.Record evidence as part of the product. Keep the source, period, revenue definition, and verification artifact together. A vibe coding app becomes a usable business case only when another reader can distinguish collected revenue from a target.

What we would refuse to build

We would refuse projects whose business case depends on volume, benchmark leakage, or a hypothetical ceiling. Those patterns can be useful experiments, but they do not justify a roadmap. A credible vibe code app needs a reachable buyer and a measurable transaction before it earns more engineering time.

• ·A mass-produced long-tail portfolio when revenue remains undisclosed [U].
• ·A directory justified mainly by another site's $1M-$5M/yr benchmark [C].
• ·A venture studio sold on theoretically uncapped profitability [C].
• ·Deep infrastructure modeled on Cursor's $500M/yr [V] before any buyer interview.

The refusal rule is simple: no additional feature work until one acquisition path produces a transaction or a specific, documented rejection.

FAQ

Yes, a vibe coding app can make money, but the useful questions are what was sold, how much was collected, and who supports the claim. The answers below use ProvenStartups classifications consistently: [V] is strongest, while [F], [C], and [U] require progressively more caution.

What are vibe coded apps?

Vibe coded apps are software products built substantially through natural-language instructions to an AI coding system, with the developer steering, testing, and correcting the output. The label describes the production method, not product quality or commercial traction. A deployed repository qualifies technically; a viable company still needs demand, distribution, and evidence.

Can a vibe coding app make money?

Yes. The full matching cohort has 22 clean monthly disclosures, with a $20K/mo median and a $2K/mo to $500K/mo range. Treat those as cohort statistics, not a promise. Payout is the cleanest compact example here: $20K/mo [V], reached in 50 days around class-action claim discovery.

Which AI coding tool should I use?

Choose the tool that lets you inspect diffs, run tests, and ship in your stack. Across the whole site, 211 distinct projects mention at least one tracked AI coding tool; ChatGPT appears in 100, Claude Code in 50, Cursor in 46, and Bolt in 40. Those are usage mentions, not success rates.

How does ProvenStartups grade revenue evidence?

[V] means third-party verified, [F] means founder-reported, [C] means creator-relayed, and [U] means unverified. The grade classifies support for the number, not whether the startup is good. That is why $30K/mo [V] for Subscribr carries a different decision weight from a $2K-$10K/mo target [C] for a directory.