How PhotoAI got its users
Built in public in front of a huge audience, with visual results that share themselves. The audience predates the product.
Confirmed. The team said it publicly, or the signals leave no room for doubt.
Reproducible, but it will cost you real money, real months, or a skill you may not have.
The edge you do not have: A personal audience built across a decade of building in public
- A launch that does not start from zero
- A direct relationship that survives algorithm changes
- 12 to 36 months of publishing before launch
- Accepting that you are the face of the product
- A topic where you have standing
Founder-led in 2026
Works extremely well, but it is an asset you build before the product, not during.
Everything on this channel →The play
5 steps, in the order they ran them.
- 1
The audience did this part. Photo AI opened with one X post from an account Levels spent a decade filling with other products shipped in public. That following predates the product, so copying the format buys nothing. What transfers is what he posted: a named scene, not a feature. "Valentina in a black hoodie during golden hour in a coworking space." With no audience, post one generated result a day with the scene written in plain words, and let the image be the whole post.
done whenLook at your last ten posts. Count how many show a finished output a stranger could judge in one second, versus an announcement about your product. If announcements win, you are talking to an audience you do not have yet.
- 2
Build the /tools wall. Photo AI hosts a grid of single-purpose free tools, each on its own URL: /background-remover, /heic-to-jpg, /image-compressor, /watermark-remover, sold as "no account, no watermarks, no catch". Every one closes with the same line pushing you to train your own model. Pick the three boring conversions your product already does as a side effect, give each a bare page and its own URL, and put that upgrade line at the bottom of all three.
done whenList the operations your product already performs as a side effect. If none of them work in one upload with no login and no email, you have nothing to give away yet and the tool pages will bounce.
- 3
Name the packs after the destination, not the model. Photo AI's packs are called Tinder, Hinge, LinkedIn headshots, AI Influencer Generator, not "portrait style 4". Each pack page is a wall of finished examples under one button, "Take photos like these". The pack is the search query and the pitch at once. List every place your users actually send the output, build one page per destination, fill it with real results, and use the destination as the page title.
done whenAsk five users where the output goes after they export it. If they name a specific place, a dating app, a job board, a client deck, that place is a page title. If nobody can name one, you are guessing at packs.
- 4
Make users publish for you. photoai.com/community is a public feed where users opt to publish their generated photos: the handle is shown, the full prompt sits in the metadata, and the page states the best published photos get prizes every week. Every entry is proof and a copyable recipe at once. Ship a publish button, expose the prompt, credit the handle, pick a winner weekly. Your customers become the poster you cannot afford to be.
done whenCheck whether one of your outputs is worth showing to a stranger with no context. If it needs a caption explaining what it is, users will not publish it and the gallery stays empty whatever prize you attach.
- 5
Turn every rival and every support reply into a URL. Photo AI runs /compare/photoai-vs-midjourney, -vs-chatgpt, -vs-runway and a couple dozen more, each a feature table naming where the rival is thin. Support answers get the same treatment: /faq/how-do-i-cancel-my-subscription plus an ID. Write one compare page per tool people already name when they describe yours, and give each repeated support answer its own indexed page instead of a help doc.
done whenRead your last twenty support messages and count the distinct questions. If the same three repeat and they live in an inbox or a private doc rather than a public URL, you are answering a search query in private.
The product itself
read from photoai.com, live“🔥 Fire your photographer”
Run this against your own product
Carries the play above, and is written to tell you no when this channel is out of reach for you.
The same twelve channels as an agent skill. It reads your repository and eliminates what you cannot run. Open source, MIT.
curl -o ~/.claude/skills/howtodistribute/SKILL.md https://howtodistribute.com/skill.md
How big it got
Revenue published openly by the founder
What else they ran
They ran the same play
Want this verdict on your product?
Describe it. You get back the channel I would run in your place, the 30-day plan, and the number at which you stop.