Voice Apps Built With ElevenLabs That Have Revenue Evidence
ProvenStartups holds 33 businesses matched by a voice-and-audio text filter across 406 projects. Only 15 name ElevenLabs at all, and only 12 name it in…
ProvenStartups holds 33 businesses matched by a voice-and-audio text filter across 406 projects. Only 15 name ElevenLabs at all, and only 12 name it in the operator's own account — the rest are voice-shaped businesses running on something else. Those 12 run from $94 in 30 days to a two-person telehealth company reporting $401M in year one.
Contents

How many apps built with ElevenLabs have revenue evidence?
Twelve, after stripping the file down. Thirty-three cases match a voice/audio text filter, 15 name ElevenLabs anywhere, one of those is ElevenLabs itself, and two name it only because our rebuild notes suggested it. Twelve operators said the word unprompted — one as a warning.
Those 12 grade out as 2 ✅ third-party verified, 4 🗣 founder-reported, 4 📎 creator-relayed and 2 🔮 unproven. ✅ means a payout screen was shown rather than described, 🗣 means the operator said it, 📎 means a channel relayed it unaudited, and 🔮 means no revenue figure exists.
One fact reframes the search: nine of the 12 are content or publishing businesses — three AI content channels, three publishing operations, three cautionary tales. On this evidence ElevenLabs is a line item in a content factory, not a business model. More teardowns sit in our built with AI hub.
What do the ElevenLabs cases actually earn?
The range is deliberately absurd: $94 at the bottom, $401M at the top, and very little in between that publishes a clean monthly number. Two figures carry ✅ grades. The rest are an operator's word or a creator's relay, which is the mix filmed teardowns produce.
| Case | Reported result | Grade | Tier |
|---|---|---|---|
| Medvi | $401M year one · $1.8B pace in year two | 📎 | Tier 3 |
| ElevenLabs (the vendor) | $125M/yr · 50 people · $2.5M per employee | ✅ | Tier 3 |
| Binge Central | $2.3M lifetime on one channel · $4M+ across five | ✅ | Tier 2 |
| Free AI workshop funnel | $400,000+ · 70% of free workshops convert | 📎 | Tier 1 |
| Barrera kids-gaming channel | $330K+ lifetime · top video $20,546 | 🗣 | Tier 2 |
| Rook finance channel | $47K in two months · peak $6,000+ in a day | ✅ | Tier 2 |
| Ghost-niche news channels | $30,000+ across four channels in 90 days | 🗣 | Tier 2 |
| Zinny Studio | ~$18K/mo · 250 members × $60/mo plus AdSense | 📎 | Tier 2 |
| AI-dubbed Spotify podcast | $500–8,000/mo, the creator's projected ramp | 📎 | Tier 3 |
| Tech Automation Insider | $94 in 30 days · ~$3/hour | 🗣 | Tier 3 |
| The 100 Club quiz funnel | $26 of ad spend returned $59 in sales | 🗣 | Tier 1 |
| ClickBank spirituality affiliate | ~100K views in 28 days · ~$50 per sale | 🔮 | Tier 3 |
| Poppy AI | Revenue undisclosed | 🔮 | Tier 1 |
Do not average this table. Medvi sells weight-loss drugs and happens to answer the phone with voice agents, ElevenLabs is the vendor rather than a customer, and Binge Central's $2.3M is lifetime. Strip those three and the starting range is $94 to $30,000 a quarter.

How did the best-documented cases earn their first dollars?
None of them sold voice. Each sold something with its own rate card — a finance channel's RPM, a drug prescription, a dental practice's missed calls — and treated synthesis as a production cost. Both ✅ cases here are media businesses whose audience existed before the voice did.
How did Rook's AI-avatar finance channel clear $47K in two months?
By aiming at the most expensive audience on YouTube rather than the largest. The American senior finance viewer carries a $24.5 RPM, so the money came from rate per view rather than output volume, with a peak above $6,000 in a single day.
We file it as a ⚠️ cautionary tale. The method clones a real person's voice with ElevenLabs or MiniMax and borrows their background, because the operator believes voice is what platforms check hardest. $47K does not price that risk.
Why does a $401M company run its support on voice agents?
Because support was the only cost that scaled with customers. Medvi reports $401M in year one on roughly $20K of starting capital and one full-time employee besides the founder. Prescribing, pharmacy and logistics are outsourced; ElevenLabs voice agents field most inbound.
The New York Times reported it in April 2026. The founder's line — you don't need to sell AI to use AI — is the sharpest thing here: the biggest case on this page is not an AI company but an old industry rebuilt underneath.
What does an ElevenLabs channel pay in its first 30 days?
$94. Tech Automation Insider published the receipts: 14 videos, 4,000 views, 103 subscribers, 30 to 40 hours of work and $280 spent on editing at $20 a video. About $3 an hour, said on camera by the person who did it.
That is the most useful number here. Every result above it rests on an existing audience, a niche rate card or years of compounding, and clean voiceover buys none of those.
Which voice businesses here do not use ElevenLabs?
Eighteen of the 33, including every serious voice-agent company here. When voice is the product the stack becomes a trade secret; when voice is a cost line, operators cheerfully name the subscription they pay for. That asymmetry is the most reliable signal in this dataset.
Sanas reports $22M+ in its first 18 months, roughly $500K per new customer 🗣. GigaML serves DoorDash and 10+ other Fortune 500s with 4–5 engineers ✅. Clarvo reports $1M ARR at $250 a seat 🗣. Setter AI wrapped a YC company's voice API and collected $500 before it had a dashboard; now $120K ARR, 40 customers 🗣. None names a vendor.
Service-tier entries are vaguer still. An AI voice receptionist runs $500–1,500/mo per client 📎; the same creator ranks the AI chat agent below it and managed AI cybersecurity above both, with no figures. A pricing framework is somewhere to start negotiating, not evidence.

What would we copy, and what would we skip?
Copy the cases where voice is a cost line inside a business that already owns an audience and a rate per view. Skip anything whose whole pitch is AI voice. The loudest dissent here comes from the operator with the biggest receipts, and it outweighs the revenue numbers.
Chris Barrera says plainly: do not use ElevenLabs. YouTube, he argues, uses Gemini and metadata to spot one recognisable AI voice reused across videos, and it already requires disclosure of synthetic content. He has $330K+ lifetime and a $20,546 video behind that warning. The nuance matters: he objects to one stock voice at volume, not to synthesis. The ClickBank case clones the creator's own voice for that reason.
Two details we would steal outright. The ghost-niche news operator pastes scripts three paragraphs at a time because longer inputs degrade delivery, and cleared $30,000+ across four channels in 90 days. The free-workshop funnel builds a voice agent for a dental practice on camera in a minute — the demo converts 70% of those rooms.
Start from the business, not the stack; if you do need it, ElevenLabs' documentation is the place. More in our built with AI hub, alongside apps built with Claude Code, apps built with n8n and vibe-coded apps.
Frequently asked questions
How much do apps built with ElevenLabs make?
Across the 12 cases where an operator names the tool, reported results run from $94 in a first month to $401M in a first year. Few publish a clean monthly figure, and the largest entries are a telehealth company and a five-year-old media portfolio. Treat the small documented numbers as your baseline.
Do I need ElevenLabs to build a voice agent business?
No, and the evidence points the other way. The four companies here that sell voice as the product — Sanas, GigaML, Clarvo and Setter AI — name no text-to-speech vendor at all. Setter AI wrapped someone else's voice API and collected $500 before it had a dashboard. Vendor choice is not the hard part of this.
What do the evidence grades mean?
✅ means a third party saw the number: a dashboard, payout screen or tracking tool shown on camera. 🗣 means the founder stated it and nobody verified it. 📎 means a creator relayed someone else's figure without auditing it. 🔮 means no revenue figure exists yet. The grade sets a claim's weight.
Is AI voice still safe to use on YouTube?
Nobody here can promise that, and two operators disagree on camera. One with $330K+ lifetime says recognisable reused AI voices are now detectable and should be avoided. Another builds his method on cloning one specific human voice to pass review. Both are betting on platform policy, which changes without notice.