Tony Dinh: TypingMind from 5-Day Build to $130K+/mo
Vietnamese indie maker who shipped a better ChatGPT UI in five days and turned it into a high-margin AI frontend business.
- MRR
- $130K–$145K
- ARR
- $1.5M+
- Users
- Enterprise + individual
- Founded
- 2023
- Team
- Small (mostly solo)
- Stack
- JavaScript, BYOK LLM APIs
How did TypingMind start?
When OpenAI released the ChatGPT API in early March 2023 I saw an opportunity. The official interface was limited. Power users wanted folders, better prompt management, multi-model support, and the ability to bring their own API keys so they only paid for actual usage. I built the first version in about five days.
I registered the domain the same day I started coding. I launched by posting on X. The first tweet got strong engagement. I added payments the same day as the first sale. Revenue came immediately: roughly $1k on day one, then $2k, then $4k. Within four days I had crossed $10k in total revenue. Within seven days it was around $22k. A Product Hunt launch a few days later pushed it further.
The early product was a better frontend for ChatGPT and other models. Users brought their own keys. That kept my costs low and gave users control. I kept adding the features power users asked for — search, organization, plugins, better context handling — and I kept posting the progress publicly.
How did you go from one-time licenses to serious recurring revenue?
Early revenue was mostly one-time license purchases. That model worked because the product delivered immediate value and because the audience was already interested in AI tools. Over time companies started using TypingMind internally. Individuals would adopt it, then ask for team features: shared workspaces, admin controls, SSO, audit logs, higher limits.
I added a Teams / enterprise plan. That shifted the revenue mix. B2B contracts, including multi-thousand seat deals, became a major part of the business. Recurring revenue from teams is more predictable and higher value than individual one-time purchases. By the time the public numbers reached the $130k–$145k per month range, team plans were responsible for a large share of that figure.
The transition required better documentation, clearer pricing, and the ability to handle security and compliance conversations. It also required saying no to custom one-off work that would not scale.
What’s your distribution engine?
X has been the primary channel from day one. I post product updates, new features, honest thoughts about AI, and occasionally numbers. The audience that followed the early launches continued to grow. Product Hunt provided spikes. The product itself generates word of mouth — people who use a better interface tend to tell colleagues.
I also run other products (DevUtils and experiments). The portfolio approach means attention and trust built on one product can transfer to the next. I do not rely on paid acquisition as the main engine. The combination of public building, useful product, and repeated launches compounds.
What does the work look like now?
I still spend a large amount of time on the product. Recent updates have included local code execution in the browser, better handling of documents and media, and memory features so the AI remembers user preferences across chats. Building with modern models is enjoyable because the capability surface keeps expanding.
Support and enterprise conversations take time. Documentation and clear product boundaries reduce the load. I keep the team small so that decisions stay fast and the culture stays close to the original indie approach.
Margins remain high because the core model is still bring-your-own-key for many users, and because software distribution costs are low. The main costs are time, infrastructure for the application layer, and the overhead of running a real business.
Lessons that still feel true
Ship the useful version fast. The first TypingMind was not perfect. It was useful enough that people opened their wallets within hours of seeing it. Charging early creates the feedback loop that polishing in private never does.
Talk about the work in public. The audience is not just marketing — it is a continuous source of feature ideas, bug reports, and social proof.
Be willing to evolve the business model. One-time licenses got the product off the ground. Teams and enterprise made it durable. Clinging to the original pricing would have left money and stability on the table.
Finally, keep building other things. A single product can be disrupted by platform changes, model improvements, or new competitors. A portfolio and an audience give you options.
More stories
Meta
The Portfolio Pattern: Why Top Indie Makers Run Multiple Products
Product Origin
Marc Lou: The Week ShipFast Changed Everything
Mature
Pieter Levels (Part 2): Nomad List & Remote OK — The Original Audience Engine
Profitable
Jon Yongfook: Bannerbear — Automated Image Generation to Solid ARR
Educator + Builder
Arvid Kahl: From FeedbackPanda Exit to Teaching Bootstrappers
Early Growth
Fernando Mendes: ScanQueue — QR Queues for Real Businesses
Scale
Danny Postma: HeadshotPro — AI Headshots to Six-Figure Months
Mature Solo
Pieter Levels: Photo AI to $100K+/mo with One PHP File
Portfolio Scale