He Built an AI Headshot Empire Solo. The Product Was Boring on Purpose.

The first thing people tell you about Danny Postma is a number. Sometimes it is the seven-figure sale of Headlime. Sometimes it is the "$300K a month" line that gets attached to HeadshotPro in every indie-hacker roundup. Both numbers float around the internet like they are the point.

I went looking for the Tuesday afternoon underneath them. What he built when the viral week ended. What he refused to build. Where the revenue actually comes from when the feed moves on.

The setup

Postma is Dutch, started young, and was not a pure engineer origin story. Landingfolio, a WordPress gallery of landing-page designs, came first when he was freelancing conversion work and needed references. Monetization was weak. The pattern that stuck was different: find a sharp job people already pay for, ship a thin product, tell the truth in public, and let distribution compound.

Headlime was the first real AI score. He had headline formulas in a book. He turned them into software. Then GPT-3 arrived and he got into an early access cohort, one of the first people wiring that model into a production copy tool alongside the early Jasper and Copy.ai wave. Public accounts of the trajectory put Headlime at roughly $1K MRR in December 2020, about $20K MRR by February 2021, and a seven-figure sale to the Jasper orbit not long after. Eight months, more or less, from product to exit.

That is the clean legend. The next chapter is the one operators should actually study.

The moment something worked, and then did not

When Stable Diffusion hit, Postma moved again. ProfilePicture.AI went out in about thirty hours. Six-figure sales in a week, per his own public telling and the interviews that followed. Fun avatars. Feed fuel. Then the curve bent. Novelty faded. Lensa-style apps ate attention. Sales cooled.

Hold on. Most founders would either double down on virality or quit the category. He listened to what people were actually asking for in the replies and support threads: not a party trick profile pic. A professional headshot. The thing you need for LinkedIn when you will not book a photographer.

He launched HeadshotPro (and, in parallel, Deep Agency, an AI "modeling agency" experiment). Deep Agency got press. HeadshotPro got the money. That split is the whole piece if you are paying attention. Hype and cash are different instruments.

The boring product

HeadshotPro is intentionally narrow. Professional headshots. Not dating packs, not meme styles, not every horizontal the competitors ship to chase App Store screenshots. Pricing, as listed publicly for long stretches, has lived in one-time packages roughly in the $29 / $39 / $59 band for batches of images, not a pure seat subscription. Refunds for unhappy customers are part of the trust pitch. The site claims on the order of hundreds of thousands of customers and tens of millions of headshots generated. Treat marketing totals as marketing totals. The direction is still clear: this is a volume utility, not a toy.

The quality claim he has repeated in interviews is the opposite of "we wrapped open source and shipped." Custom models, long pipeline work, a year of tuning toward photos that do not immediately read as synthetic. Whether you buy the "indistinguishable" language or not, the strategy is coherent: win on the job people are embarrassed to get wrong, not on feature count.

The growth engine nobody screenshots

Launch week is not why HeadshotPro kept showing up.

SEO is. Indie Hackers writeups of his approach describe programmatic pages for city and long-tail headshot queries ("professional headshots in [city]"), plus ordinary blog content, aimed at keywords with enough volume and low enough difficulty to rank without a content army. Product Hunt launches were framed, in his own comments, less as pure user acquisition and more as backlink machines: even a mediocre ranking still throws domain authority around the web. When other apps go viral for AI headshots, search interest spikes, and the site already sitting on the term catches the overflow. He has described weeks where a competitor's virality tripled his sales because he owned the search intent.

That is not a 24-hour story. That is a six-month compounding story. It also means the copyable lesson is closer to "pick a monetizable search job and build plumbing for it" than "tweet a demo."

Pieter Levels retweeting early posts is part of the lore. It is also not a strategy you can purchase. What you can copy is the public logging, the willingness to kill the press-magnet product when the sales-magnet product is obvious, and the patience for SEO timelines that do not fit a hackathon video.

The asterisks

Here is where I get careful, because the internet is not.

The "$300K MRR" figure appears constantly in secondary writeups. HeadshotPro's public pricing has often been one-time packages, not classic monthly seats, and Postma has talked in later commentary about subscription mechanics being a trap for certain product shapes. Some databases show much smaller annualized snapshots for single years. Build-in-public numbers move. Portfolio revenue gets collapsed into one headline. I am not going to pretend a blogger's roundup is an audited P&L. What I trust more is the pattern across years: early GPT product sold for seven figures, AI image products that repeatedly printed meaningful cash, a holding company (Postcrafts) still owning HeadshotPro rather than flipping it on the first offer, and a slow admission that hiring help was necessary even for someone who marketed the solo myth.

He has said he should have built a team earlier. That sentence is worth more than another revenue screenshot. The lonely version of the empire does not scale quality pipelines forever. Compute bills do not care about your personal brand.

Competition is permanent. Every model release births ten headshot clones. His answer has been focus and distribution infrastructure, not feature parity with every clone.

The verdict

If you only remember the exit, you will try to time the next GPT wrapper. If you only remember the $300K line, you will argue about someone else's spreadsheet.

The useful story is narrower. Postma keeps winning the same way: enter a technology wave early enough to matter, ship something people already budget for, follow sales rather than press, and install boring acquisition (SEO, refunds, clear packaging) that still works when the feed is bored. HeadshotPro is not clever because it is AI. It is clever because professional headshots were a real purchase before diffusion models existed, and he made the purchase faster and cheaper without turning the product into a carnival.

The product was boring on purpose. That is the part worth stealing.

Sources

• HeadshotPro, founder page
• The Bootstrapped Founder, Danny Postma interview
• Indie Hackers, HeadshotPro SEO breakdown
• Indie Hackers, Headlime exit AMA coverage
• SupaBird profile compilation (secondary; treat numbers carefully)
• Danny Postma on X