Faceless.video: $1M ARR in 10 Months — a Non-Engineer's Seventh Bubble App After Six Misses

Someone who can’t write code reaches $1,000,000 in annual revenue in ten months using only no-code tools. And on the way there, he built six products in the same environment and missed with all of them.

We read the record Jacob Seeger, formerly a composer and music producer, published on Indie Hackers.

Two and a half years to product number seven

Before Faceless.video, Seeger built six apps on Bubble, the no-code development platform, across two and a half years. None worked. Names and numbers for the six aren’t public, so they exist only at the granularity of “2.5 years, all misses.”

Number seven was Faceless.video: a service that runs faceless social-video channels automatically, scripting, editing, posting across platforms. It began crudely, layering Reddit posts read aloud over Minecraft gameplay footage.

His summary: “Six failures teach you what one success cannot. Don’t get attached.”

A similar batting average shows in 17 built, 1 hit. It is a recurring shape in our archive: behind the one that landed, a stack of ones that didn’t.

What the launch cost

Total initial outlay of $500 is essentially negligible for a business of this size. The stack is Bubble at the core, with off-the-shelf parts around it: RunPod, AWS Lambda, Replicate, Stripe, Bunny.net.

Launch paid from day one

Seeger’s refrain is about billing order: “Always launch as a paid service. The most convincing validation is simply whether someone buys.” No freemium.

Compared with free-first-then-charge, the initial user counts look worse, but the moment the first sale lands, demand is confirmed. The API developer who put up paid plans on day one takes the same order; a minority pattern in our archive, but a consistent one.

Acquisition started with the X thread, then profits were reinvested into short-form video and influencer placements up to five-figure MRR, a simple loop of buying exposure with the profit from bought exposure.

The distance between 2.5M signups and $83,000 MRR

Against 2.5 million cumulative signups, MRR is $83,000. With the best-selling plan at $35, paid users number roughly 2,000–3,000, about one per thousand signups.

That thinness is normal for generative-AI tools. The notable part is that $83,000 a month works anyway: with a large enough denominator, thin conversion still yields real money. And the denominator was built by that chain of paid exposure that began with a $250 thread.

The same “build the giant denominator first” pattern shows in the creator with 4 million followers who built an MVP in 2 days and hit $350K/month in two, audience accumulated free there, exposure bought with cash here.

How rough the first version was

Faceless.video’s initial form was crude: Minecraft footage plus a synthesized voice reading Reddit posts. Today’s pipeline, scripts through editing through multi-platform posting, accumulated from there.

So even number seven didn’t launch finished. He shipped something rough enough to build for $250, charged for it, confirmed it sold, then extended.

On architecture he offers one rule: “Build for the best case. If 100,000 people signed up today, would the app survive?” While choosing no-code Bubble, he pushes the heavy work out to RunPod’s serverless GPUs, AWS Lambda and Replicate. Owning nothing heavy is what makes sudden inflows survivable.

Position among our cases

¥12.45M a month is over 8x the median of revenue-disclosing cases here (¥1.5M), inside the top fifth, and above the SaaS-only median (¥8.41M).

No-code businesses at this level are rare in our archive. Set beside the Excel-formula AI built on Bubble during parental leave that reached $23,000/month, you can see the range of outcomes the same environment produces.

The weaknesses he names himself

Behind the good numbers, Seeger names three concerns, specifically.

Bubble’s limits: he is considering migrating to custom code but sees “many challenges.” Shipping fast on no-code and hitting its ceiling later are two faces of one choice.

Rising AI costs: inference prices are climbing and pricing needs adjustment. Video generation is especially heavy. Unit economics ride on other people’s price sheets.

Imitation: competitors building the same thing have appeared. That entry was easy for him means it is easy for others. The $250 start is also the moat’s thinness.

What this record shows

Extract only “the seventh one hit” and it’s a story about attempt counts. What actually worked here is that six failures taught him not building skills but launch order: charge from day one, buy measurable reactions, reinvest profit into the next exposure. Technically it was Bubble all along, only number seven had this loop.

Not published: profit, churn, actual paid-user counts, headcount, and the contents of the six misses. He founded alone and later built an internal team of undisclosed size. The cost structure under the “$1M ARR in 10 months” headline is not readable from this record.