The case at a glance: $5K month, 275 users, $170 MRR
- Last month revenue: $5K
- Total users, all time: 275
- MRR: $170, which he labels "WIP"
- Revenue mix: 90% one-time licenses, 10% subscriptions
- Prices: $19 / $29 one-time (bring your own API key), $5/mo for plug-and-play
- Timeline: started coding in March, launched in April
- Thread traction: 806 upvotes, 146 comments
The product reads your files with a vision model and renames them into something a human can scan: IMG_4837.jpg becomes blue-running-shoes-mens-size-10.jpg. Stack is thin on purpose — a native macOS shell, one undisclosed vision model, user-supplied API keys as the default path, nothing else named.
He came from SEO work, where he was burning afternoons renaming client images by hand. He sat on the idea for months because "file renaming" sounded too boring to be a business. Vision models got cheap and accurate at the same moment a one-person shop could ship a desktop app in a month. He almost missed the window out of embarrassment.
What he's actually selling (it isn't renamed files)
The product isn't the renamer. The product is an afternoon somebody already lost, sold back to them at $29 with no meter running.
Two mechanics do the work. First, the pain arrives in billable hours, not feelings. Nobody needs convincing that renaming 400 photos by hand is bad; the buyer already ate the cost and knows the number. That's a sale that closes in one sentence, which is the only kind of sale a solo developer can afford to make.
Second, bring-your-own-API-key quietly converts an AI product into a software product. Ninety percent of his revenue carries no variable cost — no token bill, no rate-limit firefighting, no margin that collapses when a power user goes wild. Most AI builders eat inference and call it a business. He handed the meter to the customer.
Then the distribution channel picked his customer for him. He built this for SEOs. SEOs said no, because they already have VAs, and the enemy of a $29 tool is somebody else's payroll. So he emailed Mac productivity bloggers instead — and out of that audience walked home inspectors organizing property photos by room and defect type. He didn't research his way to that ICP. He mailed his way into it.
The channel you can actually reach determines the customer you actually get.