Loading page…
A solo developer whose $14,000-a-month Twitter analytics extension was destroyed overnight by a platform price change, who sold it for $128,000 and rebuilt as several small products so no single host could do that to him again.
three products combined reported at $30,000+/month (May 2026 source) · Black Magic peaked at $14,000/month and sold to Hypefury for $128,000 in a single payment · products priced around $10–30/month · no employees, no office
Twitter repriced its API at $42,000 a month and his $14,000-a-month business was underwater the same day. He sold it for $128,000 and then built three products at once, so that no single platform could ever do that to him again.
Fewer bars = easier, cheaper, or faster for an AI-assisted solo builder. Editorial judgments based on the case details.
Tony Dinh is a solo developer from Vietnam. He was a senior engineer on a comfortable salary — one source calls it a safe six-figure job at a big tech company, the other puts it at ₩150 million a year — and he left it to ship small products in public, where the whole internet could watch him win or fail. The side project that worked first was Black Magic, a Twitter analytics browser extension. Two years took it to $14,000 a month. Then, in May 2023, Twitter repriced API access at $42,000 a month and the business was underwater immediately. He did not fight it. He sold Black Magic to Hypefury for $128,000, paid in one lump sum. What he did next is the reason this entry exists. Instead of replacing one product with one product, he built three at once: TypingMind, a faster and cleaner interface for AI chat aimed at people who live in it all day; DevUtils, a macOS utility belt for developers; and Xnapper, a macOS app that makes screenshots look good. The Korean analyst reports the three together at over $30,000 a month as of May 2026. Four months later, a second source reports TypingMind alone at over $130,000 a month, run with no employees, no office and no team, frequently from the road, and adds the detail that matters for credibility: he posts his exact numbers publicly, so anyone can check. The two figures cannot both describe the same moment, and we report both with their dates rather than picking one.
Building in public was the channel, and it was not a side activity. He shared his work on Twitter every day, which is how a Twitter analytics extension found its first users — the audience and the product occupied the same surface, so every progress post was a demo to exactly the people who would pay for it.
The quick-reference card — acquisition channels, replication playbook, risks & traps, and our verdict — is members-only.
Join the Unicorn Club — $5/moCancel anytime, in one clickPaste into Claude Code / Codex and get a working version of this product end-to-end.
Two production-grade prompts per idea — a full build spec and an SEO growth plan — plus 10 tailored build specs a week. Members only.
Join the Unicorn Club — $5/moCancel anytime, in one clickEvery claim traces back to a listed source — open any of them and check the number yourself.
Data credibility: Two independent creator breakdowns, neither a founder interview, and they disagree. A Korean solo-business channel (published May 2026) reports Black Magic peaking at $14,000/month, Twitter's API repricing at $42,000/month in May 2023, the sale to Hypefury at $128,000, and TypingMind, DevUtils and Xnapper together at over $30,000/month; it names Tony Dinh's own blog post about the sale and a Bootstrapped Founder interview as its sources. An English-language compilation (published September 2026) reports TypingMind alone at over $130,000/month and states that the founder publishes his exact numbers publicly, which is the only reason we carry a figure that far from the other. These two cannot both be current, and we deliberately report both with their dates instead of choosing. No dashboard, invoice or payment screen appears in either video. The second source's auto-caption renders the founder's name as 'Tony Den'; the correct spelling comes from the first video's own source list. The salary figure before founding is described differently by each source and we quote both. Nothing about stack, hosting costs or margins is stated anywhere; that section is explicitly our inference.
The first revenue came from a side project built while still employed, which is the structural detail worth copying. Black Magic took two years to reach $14,000 a month, and for most of that stretch the salary paid the bills and the product paid for nothing at all.
Three loops reinforce each other here. The first is the audience: building in public compounds, because every shipped feature is content and every piece of content grows the list that buys the next product.
The hard stretch here was not the beginning, it was the middle: two years to take Black Magic to $14,000 a month, and then the day the platform ended it. That second kind of silence — the business is working and then it simply is not — is the one nobody plans for.
Four products appear across the two sources. Black Magic was a browser extension that analysed Twitter accounts and their performance; it is the one that is gone. TypingMind is the current centre of gravity: a faster, cleaner, more capable interface for AI chat, built for the heavy users who live in a chat window all day and want features the official apps do not give them. DevUtils is a macOS utility for developers. Xnapper is a macOS app that turns raw screenshots into presentable images. The common shape is narrow and deliberate. One specific problem, one clean solution, as the second source puts it; the Korean analyst's version is that each of the four founders he profiled automated exactly one action and refused to expand. TypingMind does not try to be a model, a platform or an agent framework — it is the interface layer for people whose daily driver is AI chat, and its entire value is that the official interface is not good enough for the top few percent of users. The portfolio itself is part of the product decision. Three small products on three different foundations is a structural answer to the thing that killed the first one.
The stated price band across his products is roughly $10 to $30 a month, which the Korean analyst identifies as the common range across all four solo founders in his breakdown — low enough that the purchase needs no justification, paid for long enough that the compounding does the work. The reported revenue numbers are the disputed part and we present them as they were given. In May 2026, three products together were put at over $30,000 a month. In September 2026, TypingMind alone was put at over $130,000 a month, with the source noting that the founder publishes his exact figures publicly. Both are creator reports about a third party; they cannot both be current. The exit is the cleanest number in the story: Black Magic sold to Hypefury for $128,000 in a single payment, after peaking at $14,000 a month. That is roughly nine months of revenue as a multiple, which is what a distressed sale looks like when the platform has already decided your costs. Our read: the portfolio structure changes the shape of the income rather than just its size. Three products at $10,000 a month each is a far more durable business for one person than one product at $30,000, because no single host, store policy or API price can take the whole thing.
Neither source describes the infrastructure. What both establish is the headcount: no employees, no office, no team, one laptop, frequently on the move. Our read: the cost structure splits sharply by product type. DevUtils and Xnapper are native macOS apps — near-zero marginal cost, a licence server and a download, margins limited mainly by Apple's cut where the store is used. TypingMind is the one with a real variable-cost question, because a front-end for AI chat either passes the model bill to the user, by having them supply their own API key, or absorbs it and resells inference. Those are two completely different businesses: the first has software margins and a technical audience; the second has to price above a cost that the model vendors move without asking. Neither source states which applies, and anyone copying this needs to decide it deliberately on day one. The cost that is certain is maintenance attention. Four products by one person means every platform change — a macOS release, a browser extension migration, a model vendor's API revision — lands on the same person's desk, and they arrive on schedules he does not control.
Copy the diversification, and copy it before you need it. The single most transferable decision in this story is that after one platform ended one business, he did not go looking for a better single platform — he built several small products with different dependencies. The Korean analyst states the rule plainly: do not bet your life on one product, spread the platform dependency, build small things you personally use every day. Copy the public numbers too. Publishing exact revenue is what turned a solo developer into a case study that two unrelated channels in two languages built segments around, at zero cost to him. Change the audience assumption. The build-in-public flywheel that worked here was earned over years on a platform that has since changed its economics and its reach; anyone starting now is building that audience somewhere else, more slowly, against more noise. Our read: the harder thing to copy is the range. Shipping a browser extension, two native macOS apps and an AI interface alone requires genuine breadth, and AI assistance narrows that gap without closing it. The realistic version for most people is two products, not four, with the second one started only once the first can pay rent.
This entry is its own cautionary tale, which is what makes it useful. A product at $14,000 a month became unprofitable in a single announcement when Twitter moved API access to $42,000 a month. No product decision caused it and no product decision could fix it. TypingMind carries a different version of the same exposure. It sits between users and the model vendors, and those vendors ship interface improvements constantly — every feature the official apps absorb removes part of the reason to pay a third party. API pricing and terms can also change underneath it, exactly as Twitter's did. The category is crowded with competitors doing the same thing, several of them free, and the switching cost for a chat interface is close to zero. The native apps carry platform risk of their own: an operating system release can break a utility, and store policy can remove distribution. Our read: the portfolio is the mitigation, and it is a real one, but it is not a moat. What actually protects this business is speed — one person who ships faster than the vendors absorb features, sustained for years. That is a personal characteristic, not a defensible asset, and it is the reason to read this as a case study in structuring risk rather than as a product to clone.
The four founder playbooks — first customers, first dollar, the flywheel, the silent stretch — are in the Unicorn Club, on this idea and on all 1,435 others. The story, the numbers, the sources and every later section stay open to everyone.
Join the Unicorn Club — $5/moCancel anytime, in one click