Hello everyone! 👋 It’s Tony again 😄

March 2023 has been a crazy month for me, both positively and negatively. Here’s a recap:

• My new AI product typingmind.com was a success!

• Twitter announced “no more free API” and charged $42K/month. What will happen to Black Magic?

Let’s go into the details!

My new product TypingMind.com

At the beginning of the month, I started working on a new product.

I’ve been using ChatGPT heavily recently, and I started to find out there are a lot of inconveniences with the default chat user interface OpenAI has. For example:

• I have to log in again every day (very annoying!)

• ChatGPT types out the answer very slowly.

• There is no way to search past conversations.

There were a lot more pain points, but those two were the most annoying for me.

Then, on 1st March, OpenAI announced ChatGPT API.

Right on that day, I came up with the idea: I must create a new UI to solve my own pains!

On 2nd March, I registered the domain name typingmind.com (suggested by ChatGPT 😂)

Then, I started working on the product.

Over the next few days, I worked on typingmind.com on and off. It started very slowly. I didn’t have a lot of motivation and started to get distracted.

Then the weekend came, and I decided to put all of my weekend into finishing it. Otherwise, I may never finish it.

On Monday, 6th March, 5 days after OpenAI announced their API, I finished the first version and shared it on Twitter:

Tony Dinh 🎯 @tdinh_me

My weekend project is now available: typingmind.com I was so annoyed by how ChatGPT slowly typed out the answers + kept logging me out every day, so I built a new UI on top of its API. (plus some more convenience features) Free to use with your own OpenAI API key 😄

typingmind.comTypingMind — A better UI for ChatGPTUse ChatGPT with enhanced features like chat history search, folders, integrations, prompt library, etc....

6:38 AM ∙ Mar 6, 2023

---

990Likes100Retweets

The app got a lot of attention, so I decided to add payment. Then I get my first sale:

Tony Dinh 🎯 @tdinh_me

First sale on @lmsqueezy 🥳🥳🥳

3:46 PM ∙ Mar 6, 2023

---

109Likes3Retweets

That feeling of the first sale never gets old!

I was so excited!

In the following days, I continuously added more and more features.

Every time I add a new feature, I tweet about it, and more sales come in!

Within 1 day, I hit $1K in revenue.

One more day, $2K.

One more day, $4K.

By 10th March, I hit $10K in revenue from my weekend project.

99% of the sales are from my Twitter reach.

Tony Dinh 🎯 @tdinh_me

$10K revenue! 🥳 - started as a weekend side project - domain typingmind.com bought on 2nd Mar - MVP released on 6 Mar - 7 Mar: $1K - 8 Mar: $2K - 9 Mar: $4K - 10 Mar: $10K! - 11 Mar (tomorrow): Launch on Product Hunt!

3:49 PM ∙ Mar 10, 2023

---

1,828Likes93Retweets

I worked on the project like crazy. Added new features, improvements, and bug fixes every single day, multiple times a day.

Every time I add a new feature, I tweet about it, and that’s my entire marketing strategy.

You can see a timeline of all of my tweets about the features (that I now use as the changelog) on this Notion page.

People love the product because:

• It was faster than ChatGPT at the time (*)

• A lot of convenience features (search, define character, prompt library, etc.)

• People can use their API key, paying only for what they use. No monthly fee!

• And most importantly (in my opinion): people don’t have to re-login every single day 😂

(*) At the time of writing this, OpenAI has already improved the speed, but people have already fallen in love with TypingMind for various other unique features!

On 11 March, I launched the product on Product Hunt.

With a lot of help from my audience on Twitter and in this newsletter, TypingMind got to #1 product of the day.

Thank you all so much if you were there on the launch day! I really appreciate your help.

Revenue went from $10K → $22K in a day.

Tony Dinh 🎯 @tdinh_me

Launch report for typingmind.com! 🥳 🏆 #1 Product of the Day ⬆️ 902 votes 🎉 Added ~$5K revenue 💰 $22.7K total revenue Thank you all!!! This is the beginning! I have a 2-week plan of what's coming next. Gonna reveal in my next tweets!

9:02 AM ∙ Mar 12, 2023

---

748Likes17Retweets

I was extremely surprised by the result.

When I made my first sale, I thought, "Okay, perhaps I could earn a few thousand dollars here."

Making $22K is sooooo out of my expectation.

I totally underestimated the hype around AI.

Obviously, I continue to make more revenue after the Product Hunt launch. But I decided to stop sharing the revenue after the launch for various reasons.

Did I get lucky?

Reading the timeline above may make you think that either:

• it’s very easy, or…

• I’ve been very lucky, or…

• I made it because I have ~76K followers on Twitter.

All of those may be true. There is no way I can tell what made it work.

But here are some facts that may help you better understand the context.

Firstly, TypingMind was not my first AI product.

I got hyped with the whole AI thing long ago (I mentioned that in my past newsletter issue).

I tried to build two different AI products before: EmojiAI and AskCommand, but both didn’t work out.

When I started those two products, I had ~50K followers on Twitter.

Especially for AskCommand, this promote tweet got ~5,000 likes. One would think the product would be so successful.

Nope, I ended up only making ~$100 for both attempts.

In the end, I think having a big audience helps, but it will not guarantee success. And I definitely will not make any meaningful money making a todo list 😂

These comments from Glavin and Jason sum it up very nicely:

Secondly, I have some unfair advantages.

I already talked about this in my December 2021 newsletter, so I won’t repeat myself here. In short, my unfair advantages are:

• My audience (Twitter, newsletter): I spent a lot of time and effort over the past 2 years building an audience. I know I’m not good at marketing, so building an audience is the best way for me to have a distribution channel.

• My technical skills: I can write code very fast. Coding is the easy part for me. I’ve been writing code since 15. I have written code almost every day over the past 8 years. I can’t solve medium Leetcode problems and don’t write the best clean code out there, but I still love writing code and putting code together to make something useful.

• Some “product sense”: By going solo and doing everything by myself, I save a lot of time collaborating and communicating. I know which features to cut and which tech debt to take to provide value to customers quickly and make them happy.

I encourage everyone to find and build some unfair advantages for themself!

Things I did differently this time

I want to share some thoughts on why TypingMind was a success.

First thing is that I took another approach to the pricing model.

TypingMind is a one-time purchase software.

It’s tempting to build just another wrapper using OpenAI’s API, charge people a monthly fee, and earn that sweet recurring revenue.

But think about it: everyone is doing that. If I were to do the same way, the competition would be huge. Platform risk will also be huge.

Jon Yongfook @yongfook

AI startups building on GPT realizing that new GPT basically replaces their product offering.

1:36 AM ∙ Mar 15, 2023

---

4,117Likes307Retweets

OpenAI makes it so easy that eventually, people are going to add whatever your product offers to their product “as a feature” using OpenAI’s API directly anyway (or at least that’s what I predict).

So what I did was a tool, just a good old regular tool, in the form of a static web app.

No backend, no server, and no database.

People bring their own API keys (which OpenAI is ok with if you only store the API locally on the user’s device).

Just like Postman or another HTTP client app, where you use the software to send requests to a server. TypingMind is a Chat client that can connect to OpenAI’s API for you and provide you will a lot of convenience features to improve the chat experience.

I even offer a self-hosted version of TypingMind that the customers can run from their local device or upload it to their own server/domain.

It’s just a static web app.

And because I don’t have any server or database to maintain, there is no recurring cost.

That’s why I can afford to (and decided to) charge a one-time purchase.

I think that decision contributed a lot to the success!

What’s next for TypingMind:

I recently brought my freelancer to start working on TypingMind with me.

It’s no longer a one-man show 😄

This will help speed up the development of TypingMind even more.

My ultimate plan for TypingMind is simple: become the best AI Chat UX.

Here are what I’m planning:

• Adding more useful features, of course.

• Working on bringing Custom Deployment for teams, enterprises, and organizations. I’m hoping to step my foot into the B2B market! Hope it works.

• Support other LLM Chatbots (e.g., Claude)

• Somehow integrate with lang-chain for maximum power!

• Keep up with the new features and capabilities from OpenAI, like ChatGPT Plugins.

I’ll share more about the plan in my next newsletter issue. Can’t wait!

Now, for the sad news

That was all I have to share about TypingMind, it was all fun and games.

Now the sad news: Twitter announced that they are going to stop the current free API access.

Their new minimum price is $42,000 per month. Yep, you read it right.

They also announced 2 other API pricing plan: free and $100/month, but both of these plans are so limited that I think nobody can build any serious business with it.

My Twitter product Black Magic only makes ~$16K MRR at the moment.

Over the past few days, I’ve been thinking about this and concluded that the only way to move forward is to merge Black Magic into a bigger Twitter product that can afford to pay the $42K/month price.

I’m in an acquisition process. All I can share right now is:

• Black Magic will continue business as usual

• I still continue to work on Black Magic

I’m happy that the customers can continue to use Black Magic, and I can continue working on it. I also think the new owner is a perfect fit for all the existing customers, and they’ll like it.

I’ll share more about this in an official announcement soon!

That’s all for now!

I spent an entire morning writing this newsletter (instead of shipping new features).

If you like it, please share it with your friends! 😄

With platform risk becoming more real these days, I really hope to get my audience to this newsletter as much as possible.

Unlike on Twitter, where Elon can ban me anytime (or even Twitter will disappear?). This newsletter is where I can control my own destiny.

Thank you for your help!

See you next time 👊

- Tony