46 Apps Built With Cursor That Have Revenue Evidence

ProvenStartups tracks 46 apps built with Cursor that have revenue evidence, and 29 are solo-run. Across the full matching cohort, 13 publish a clean monthly figure: the median is $14K/mo, with cases ranging from AEO Service at $2,000/mo [F] to Gravl at $440K/mo [V]. Our read is simple: use Cursor to compress implementation time, but refuse to treat faster coding as proof of demand.

Contents

• ·What the 46-project cohort shows
• ·Revenue comparison by product and evidence
• ·What we would build and refuse to build
• ·Where the data contradicts the Cursor narrative
• ·A practical Cursor validation plan
• ·FAQ

What the 46-project cohort shows

The 46-project cohort is useful because it separates a real pattern from demo culture: Cursor appears across recurring SaaS, consumer apps, directories, plugins, and scale references, not one magic business model. The strongest signal is breadth plus solo execution, while the weak signal is any claim that Cursor itself created demand.

The $14K/mo median is calculated over the full set of 13 cases with clean monthly figures, not only the named samples below. The range is wide enough to make the median a screening reference, not a forecast: $2,000/mo [F] for a single-client AEO retainer at one end and $440K/mo [V] for Gravl at the other.

The category mix is concrete:

• ·15 SaaS and 9 consumer apps
• ·5 scale references, 4 AI websites, and 4 platform plugins
• ·3 directory sites, 2 simple tools, and 2 cautionary tales
• ·1 ecosystem tool and 1 AI service

That mix matters because only 2 projects are classified as simple tools. ProvenStartups keeps revenue source separate from revenue size; the grading method explains why $15K MRR [F] is a different claim from $13K MRR [V].

Revenue comparison by product and evidence

Use the table as a map of business shape, not a leaderboard. A $600K+/mo [F] founder claim and a $34,000–$35,000/mo [V] third-party-verified figure are not interchangeable, even when both products were built with Cursor. Revenue size, source quality, category, and implementation difficulty all change what the case can teach.

[V] is third-party verified; [F] is founder-reported. The grade does not declare a founder dishonest. It tells you how much independent support sits behind the number, so a spectacular claim such as Minea at $750K MRR [F] does not silently outrank SiteGPT at $13K MRR [V].

What we would build and refuse to build

We would build a narrow paid workflow with an obvious buyer, a short path to value, and a distribution surface that exists before the code. We would refuse a generic AI wrapper whose only advantage is that Cursor made it quick to ship. Build speed reduces cost; it does not supply urgency, trust, or acquisition.

Three patterns are worth copying:

1. 1.Operational SaaS: Setter AI reached about $10K MRR [F] at difficulty 2/5. The useful pattern is a specific commercial job, appointment setting, rather than an open-ended assistant.
2. 2.Product plus acquisition asset: SiteGPT reports $13K MRR [V] and used free-tool SEO, while Cursor Directory reports $34,000–$35,000/mo [V]. Distribution is part of the product design.
3. 3.Consumer utility with lean economics: HabitKit reports $15K MRR [F] with only $200–$300/mo [F] in costs. The lesson is recurring utility and controlled overhead, not merely a polished interface.

We would reject “ship in days, add a $99 price, then wait” as a strategy. The four-day AI app case published $12K in its first month [U], but the blueprint is a methodology, not a verified repeatable product. That distinction is exactly where most Cursor roundups become useless.

Where the data contradicts the Cursor narrative

Popular Cursor content says the opportunity is easy weekend micro-SaaS. This cohort contradicts it: only 2 of the 46 projects are simple tools, while 15 are SaaS, 9 are consumer apps, and 5 are scale references. Cursor is common in serious products; it does not make serious products simple.

The revenue leaders make the contradiction sharper. Revid reports $600K+/mo [F] at difficulty 4/5, while Cursor itself is a difficulty 5/5 scale reference at $500M/yr [V]. Those are not “prompt, deploy, collect passive income” businesses.

Solo operation is common, with 29 of 46 projects run by one person, but solo does not mean effortless. The better inference is that Cursor can increase one developer’s implementation capacity. It cannot replace positioning, support, distribution, or the evidence needed to trust a revenue claim.

A practical Cursor validation plan

The practical move is to validate the paid workflow before expanding the build. Use Cursor for the smallest end-to-end path that reaches a chargeable outcome, then let observed buyer behavior decide the backlog. A founder who cannot name the buyer and acquisition channel should not start by generating more features.

1. 1.Choose one expensive job. Write the buyer, trigger, input, and paid result in four lines.
2. 2.Inspect the implementation boundary. Use Cursor’s official documentation for product behavior and the Cursor engineering blog for technical context.
3. 3.Build one complete path. Include authentication, the core action, payment, failure handling, and a manual fallback.
4. 4.Charge before broadening. AEO Service’s single-client retainer is $2,000/mo [F]. One paying workflow is better evidence than a crowded waitlist.
5. 5.Record the claim correctly. Save the source, period, gross-versus-net basis, and evidence class next to every revenue number.

Do not set Revid’s $600K+/mo [F] as a launch target. Use the cohort median of $14K/mo as context only, then optimize for the next verified transaction. The metric that matters first is whether a defined buyer pays for a defined result.

FAQ

These cases show that Cursor can support real revenue-producing software, not that installing it creates a business. The useful questions concern evidence quality, solo feasibility, product selection, and where to inspect the underlying records. Treat every figure as a sourced observation with a grade, never as a promised outcome.

What does “revenue evidence” mean here?

It means a revenue claim is preserved with its source class. [V] is third-party verified, [F] is founder-reported, [C] is creator-relayed, and [U] is unverified. A $500M/yr [V] figure for Cursor therefore carries different support from a $600K+/mo [F] figure for Revid, even though both remain useful cases.

Can a solo founder build a viable app with Cursor?

Yes, but the data supports feasibility, not certainty: 29 of the 46 matching projects are solo-run. HabitKit is a useful reference because it reports $15K MRR [F] and $200–$300/mo [F] in costs. Product scope and distribution still determine whether one operator can sustain the business.

Which Cursor business model would we choose first?

We would choose narrow B2B workflow SaaS at difficulty 1/5 or 2/5, with a buyer reachable before launch. AEO Service at $2,000/mo [F] and Setter AI at about $10K MRR [F] show two versions: a focused retainer and a repeatable product. We would not begin with a mass-market clone.

Where can I review more cases?

Browse the full ProvenStartups index, which contains 406 graded startup ideas, including 266 software or SaaS products and 38 cautionary tales. Start with cases matching your buyer, distribution channel, and acceptable difficulty. Compare evidence grades before comparing revenue, because the largest number is not automatically the strongest evidence.