40 Apps Built With Bolt.new That Have Revenue Evidence
There are 40 apps built with Bolt.new in the ProvenStartups cohort, and 25 are solo-run. Only 10 publish a clean monthly figure; across the full matching…
There are 40 apps built with Bolt.new in the ProvenStartups cohort, and 25 are solo-run. Only 10 publish a clean monthly figure; across the full matching set, the median is $67K/mo and the range is $37/mo to $2M/mo, with no single evidence grade for that mixed aggregate. The straight answer: Bolt can compress implementation, but distribution and proof still separate a startup from a generated demo.
Contents
Use the cohort summary to see what exists, the evidence table to inspect the strongest numbers, and the contradiction section to avoid the usual vibe-coding trap. The final sections turn those findings into a narrow build filter and a validation sequence suitable for a developer working alone.
What the 40-project cohort shows
The full matching cohort contains 40 projects, not a hand-picked set of winners. SaaS is the largest category with 9 cases, followed by 7 consumer apps and 5 AI services; the set also includes 5 cautionary tales. That mix matters more than a gallery of polished screenshots because it preserves failed business models beside working ones.
Solo operation is common: 25 of the 40 projects are run by one person. It is not proof that every product was trivial to build. The cohort spans difficulty ratings from lightweight implementations to infrastructure-scale products, while the clean monthly revenue subset contains only 10 cases.
There is also a data limitation worth stating plainly. The cohort summary’s evidence split records 5 cases as [V] third-party verified and zero as [F], [C], or [U]; it does not classify the remaining cases in that summary field. ProvenStartups does not silently fill that gap. Individual records retain their disclosed grade.
For context, the full ProvenStartups index contains 406 ideas, including 266 software or SaaS products, 246 solo-run businesses, and 38 documented cautionary tales. Bolt appears in 40 cases across a site-wide set of 211 distinct projects that mention at least one AI coding tool.
Revenue evidence, strongest cases first
Start with verified outcomes, then lower confidence as the evidence weakens. The table does not claim Bolt alone created the revenue, valuation, users, or exit. It shows projects in the Bolt cohort and preserves the source grade beside every commercial figure, which is the only honest basis for comparing spectacular outcomes with repeatable solo opportunities.
| Case | Disclosed result | Grade | Category | Difficulty |
|---|---|---|---|---|
| The Viral App Monetization Machine | Cal AI and Lerna: $2M/mo each [V]; LazyFit, CoinSnap, and Impulse: $700K/mo each [V] | [V] | Consumer App | 5/5 |
| Influencer-Partnered App Venture Studio | About $67K/mo [V], or roughly $800K ARR [V], from Scam Profit | [V] | AI Service | 4/5 |
| Base44 | $80M exit [V]; $1M ARR in 3 weeks [V]; about 400K users [V] | [V] | Scale Reference | 5/5 |
| Replit | $3B valuation [V]; about $160M ARR [V]; 350K paid apps [V] | [V] | Scale Reference | 5/5 |
| The 4-Day AI App pricing blueprint | $12K in the case study’s first month [U] | [U] | AI Website | 2/5 |
| Unblocked Games Site | $15K/mo [C]; site sold for $120K [C] | [C] | AI Website | 2/5 |
Lower-grade records can still expose a model, but they cannot carry the same conclusion. AI Book Writing claims $2.2M over 3 years [U], about $8 per book [U], and an estimated $500/mo [U] at a stated rank. The AI Chat Agent describes a monthly retainer [C], but no amount was disclosed.
That distinction is the product, not fine print. Read the ProvenStartups grading method before using any case as a benchmark: [V] is third-party verified, [F] is founder-reported, [C] is creator-relayed, and [U] is unverified.
What the data contradicts
Revenue data contradicts the popular claim that faster app generation makes software businesses easy. The cohort is solo-heavy, yet only 10 of 40 projects disclose a clean monthly figure, and 5 are cautionary tales. Fast implementation expands the number of attempts; it does not create acquisition, retention, payment intent, or trustworthy evidence.
The strongest counterexample is hiding in plain sight. The consumer-app winners reach $2M/mo each [V], but that case carries difficulty 5/5. Base44 reached a $80M exit [V], also at difficulty 5/5. The largest outcomes are not evidence that the whole business became a weekend prompt.
We would therefore refuse the default “ship first, find users later” advice. The bottleneck is usually a repeatable distribution loop and a paid problem. Bolt reduces the cost of testing that loop; it does not remove the loop.
What we would build and refuse to build
We would build a narrow product attached to distribution that already exists: a consumer workflow with a measurable paywall, a service-backed tool with reachable buyers, or a utility distributed through a partner. We would refuse generic AI wrappers, traffic-dependent clones, and automation schemes whose revenue story is stronger than their evidence grade.
Build candidates:
- ·A consumer app where activation, paywall exposure, conversion, and renewal can be instrumented from the first release.
- ·A service-backed AI tool sold to a defined buyer. The AI Chat Agent has no disclosed number [C], so treat it as a validation pattern, not a revenue benchmark.
- ·A partner-distributed product. The influencer-led venture studio’s roughly $67K/mo [V] shows why embedded reach can matter more than another feature.
Refuse or heavily constrain:
- ·A broad “AI for everyone” wrapper with no owned channel, switching cost, or specific paid event.
- ·An ad-only clone. The unblocked-games case reached $15K/mo [C] and sold for $120K [C], but creator-relayed evidence should not become a forecast.
- ·“Automated” commerce without operational modeling. Semi-Automated AI Cross-Border E-commerce reports about RMB 37K/mo profit [F] on RMB 400K+/mo GMV [F] and is filed as a cautionary tale.
The filter is blunt by design: if distribution cannot be named before the scaffold, do not build the scaffold.
A validation sequence for Bolt.new founders
Validate the paid event before polishing the generated app. Pick one buyer, one painful workflow, one acquisition channel, and one observable conversion. Use Bolt for the smallest end-to-end test, then collect evidence in increasing order of strength. Product speed is useful only when it shortens the distance to a falsifiable business result.
- 1.Write the buyer, problem, channel, and paid action in one sentence.
- 2.Build only the path from arrival to completed outcome and payment.
- 3.Use Bolt.new’s official help center for product-specific implementation questions.
- 4.Instrument activation, paywall views, purchases, refunds, and renewal behavior.
- 5.Record the source behind every claim and assign the appropriate evidence grade.
Do not turn the 4-day blueprint’s $12K first month [U] into an expected value. Treat it as one unverified case, then demand stronger proof from your own product before expanding scope.
FAQ
The practical conclusion is narrow: apps built with Bolt.new can become real businesses, but the builder is not the business model. Use the cohort to choose patterns, grades to control confidence, and a paid validation loop to decide whether a prototype deserves more engineering rather than assuming speed is evidence.
How many revenue-backed Bolt.new projects are in the cohort?
The matching cohort contains 40 projects, with 25 run solo. Ten disclose a clean monthly figure; over that full matching subset, the median is $67K/mo and the range is $37/mo to $2M/mo. Those aggregate values mix underlying records, so ProvenStartups does not attach a single evidence grade to the median or range.
How trustworthy are the revenue figures?
Trust depends on the grade beside each figure, not its size. Across the full 406-case index, 57 records are [V], 184 are [F], 121 are [C], and 44 are [U]. A $2M/mo result [V] can support a stronger conclusion than a dramatic creator-relayed claim, though neither proves a new app will reproduce it.
What type of Bolt.new app should a solo founder build?
Build the narrowest product with a reachable buyer and observable payment event. The cohort’s largest buckets are 9 SaaS cases, 7 consumer apps, and 5 AI services, but category frequency is not a recommendation by itself. Prefer an existing channel, measurable retention, and limited operational load over whatever demo looks most impressive.
Does Bolt.new turn an app into a startup?
No. Bolt can help produce and iterate software, while a startup still needs a scalable commercial model. Replit’s roughly $160M ARR [V] and Base44’s $80M exit [V] are scale references, not typical prototype outcomes. The defensible takeaway is faster experimentation, not automatic demand, revenue, or company formation.