From DevUtils to TypingMind, Tony Dinh Chose a Different Path for Each Product

Earning a First Sale from a Small Mac App

In 2020, Tony Dinh, then working as a software engineer in Singapore, started building a Mac log viewer and stopped after six months. He had set out to have a good-looking interface, many features, and a refined structure, and the project kept growing without end. With his next attempt, DevUtils, he finished the first version in about two weeks. He made use of the Mac app development experience he had gained from the earlier failure, but narrowed the scope of the first product to be released.

DevUtils was a Mac app that gathered the data conversion and inspection tools developers use often into one place. It reduced the trouble of hunting for websites every time he wanted to format JSON neatly or decode an encoded string, and he did not have to paste data into outside sites. It also offered a way of working where content copied to the clipboard could be called up with a keyboard shortcut and processed right away. The value Dinh sold lay in the convenience of finishing repetitive work quickly and in local processing.

After distributing it free to friends and colleagues and confirming a good response, he attached a one-time purchase of $9. In September 2020 he unveiled the product on Hacker News and experienced his first sale. At that time he published the source code and charged for an app that users could install without having to build it themselves.

The problem was that the interest right after launch did not last long. After visitors flooded in from Hacker News and Product Hunt and sales went quiet, Dinh began looking for a way to keep meeting new customers. He changed the approach of only repeating product promotion on Twitter and made public the small experiments he built himself, his development process, and conversations with other developers. He was starting to build an audience he could tell whenever he released a product.

BlackMagic's Growth and Going Independent

His steady use of Twitter led to a second product. Around May 2021, with 1,000 followers close at hand, Dinh wrote code that displayed goal progress around a profile picture. When people showed interest, he turned it into a web service and charged $4 a month for paid features such as customizing colors. This was the start of BlackMagic.

In August 2021, BlackMagic's monthly recurring revenue was $318. Around then Dinh decided to wrap up seven years of working as an employee developer and go independent. He already had products on sale and savings that could cover more than two years of living costs, and he said he planned to focus on DevUtils and BlackMagic for the time being. Even with small early revenue, he had secured the time to improve his products and try other things.

The turning point in BlackMagic's growth was Magic Sidebar, a browser extension that attaches to the Twitter interface. It let users check which conversations they had had with a particular person in the past, look at how a post's response changed over time, and leave notes on an account. Dinh gathered next to the existing interface the features he himself had needed while running Twitter. To users who had come in to decorate their profile, he presented a new use that helped them run their account.

Development and sales took place with the same people. Dinh built a feature in the morning, published the deployment and a demo video in the afternoon, and spent the night talking with test users and fixing errors. About 920 people took part in an invite-based trial over roughly 60 days, and the results he published in December 2021 were 248 paying users and $2,164 in monthly recurring revenue. The more he improved features, the more there was to introduce, and the people who saw those introductions took part in improving the product again.

The Crisis That Came While Testing a New Product

As income grew, he tested new ideas while running his existing products. In June 2022 he released a trial version of Xnapper, a Mac app that tidies up screen captures. The starting point was being asked repeatedly how the images he posted on Twitter had been made, and he sold $7,802 worth of licenses through presales. It was a case of applying the Mac app development experience he already had to demand he had discovered through his existing activity.

But having followers did not mean every new product sold. EmojiAI and AskCommand, the AI products he made before TypingMind, fell short of the results he expected. He had about 50,000 followers at the time and a promotional post for AskCommand drew about 5,000 likes, yet the revenue from the two attempts combined was about $100, Dinh said. There was still a gap between the ability to attract attention and the ability to create a reason to buy.

In early 2023, Twitter began charging for the API that provides data to outside apps. Dinh was told that the enterprise API pricing BlackMagic needed started at $42,000 a month. At the time BlackMagic's monthly recurring revenue was about $14,000. Even if customers kept subscribing, it was a situation where he could not cover the new API cost.

Meanwhile another opportunity opened up. When OpenAI released the ChatGPT API on March 1, 2023, Dinh decided to improve himself the chat interface he found inconvenient. The next day he registered the TypingMind domain, finished a first version over the weekend, and unveiled it on Twitter on March 6. He took as his starting point the experience of having trouble finding past conversations and having to log in again so often.

TypingMind's Early Sales and a Business Reshuffle

TypingMind's early personal product worked by having users connect their own API key. Dinh provided a convenient chat interface and sold a software license, while users covered model usage fees from their own API accounts. He also offered a structure that could be used in a local browser or a self-hosted environment. It was a design that separated the software sale price from model usage fees that grow with usage.

This time the buying response was different. Dinh announced each added feature on Twitter, and reported that after a Product Hunt launch he earned about $22,000 in license revenue over the first seven days. The audience he had already built contributed to the early sales, but compared with the failure of the two earlier AI products, follower count alone does not explain the result. Real purchases gathered around a product that solved the specific inconvenience of a service people used often.

BlackMagic was eventually sold to Hypefury for a total of $128,000. Dinh had turned down a $500,000 acquisition offer just a few months earlier, but in a situation where API access could be cut off, he had less time and fewer options to negotiate. He found a way to continue the service by folding the product into Hypefury's structure, which could bear the enterprise API contract.

At this time Dinh judged that even if he lost BlackMagic he could get by on income from his other products. He actually considered shutting the service down instead of going through a sale, but chose the side where existing customers could keep using it. Because income came from several products, he still had room to choose even when one business wavered.

As TypingMind grew, the way he operated changed too. In June 2023 it brought in about $30,000 in monthly revenue, with one full-time person handling customer support, marketing, and content and two part-time developers. The developers also looked at customer inquiries themselves and resolved errors, and work was mostly coordinated through asynchronous messages. He began handling with a small team the problem that the founder's repetitive work grows along with the product.

He used outside services for payments and license management. TypingMind used Lemon Squeezy for payments, sales tax handling, license key management, and affiliate sales features. If you implement yourself every operational function each product needs, the burden of maintaining several businesses grows along with it. Dinh's team used existing services for this common work and allocated time to product development.

A Different Choice for Each Product

In March 2024 he sold Xnapper for $150,000. At the time Xnapper was generating about $4,000 a month in revenue, but Dinh was spending more time on TypingMind, which earned more than $40,000 a month. He judged that he was not making the most of Xnapper's growth potential. By finding a buyer and handing the product over, he reduced the number of products he had to run.

The Xnapper deal also reflected the limits of a founder's personal audience. The payment was set up so that $100,000 came first and the remaining $50,000 was received as a share of later revenue. Dinh explained that, considering the influence his Twitter account had on sales, this was a condition that gave the buyer time to establish other marketing channels. The attention generated by a personal account could not be transferred intact along with the product.

He kept DevUtils. It already had a substantial portion of the features needed and took little effort to maintain, so it could keep selling steadily even while he focused on TypingMind. TypingMind, by contrast, kept opening up more room for customer requests and new features. Dinh weighed each product's growth potential and the time it required, and made different choices about keeping, selling, and focusing.

The results of that focus grew as he expanded a personal app into an enterprise product. In November 2024 Dinh said TypingMind had recorded $1 million in revenue over the previous 12 months. For the team product that businesses use to set up internal AI services, he also signed a contract covering 3,000 users and several contracts covering 1,000 users. A path was also created where some of the people who encountered the product through a personal license went on to an enterprise subscription.

In October 2025, TypingMind's published monthly revenue was about $130,000 to $160,000, and recurring revenue from the enterprise team product was more than half of total monthly revenue. One-time purchases and subscriptions ended up serving different customers within the same product. Dinh, who had increased his income sources by building many products, expanded the revenue structure within the promising product he had found.

In June 2026, too, Dinh released a new small mobile app while saying he would keep TypingMind as his main focus. The experiment of building a mobile app with an AI agent led to a plan to provide more tools and a runtime to TypingMind as well. As exploring the next product came back as experience that improved his main product, the activity of building many products and the activity of focusing on one business continued together.