Thank you for reading this month’s newsletter. Last month, I wrote that I wanted to focus on one or two projects in the long run. Well, I went off script and started working on a new project - read further for more details on how and why.

Puzzle AI - Second App

Not my idea

It all started with the following picture I saw on Twitter. I was pretty impressed by the illusion generated by AI.

Then I saw another tweet by Ash, where the above image was used in a puzzle.

I thought this was a great idea for an iPhone Game. Instead of the tiles which you switched, I wanted a sliding game.

The idea would be to have one of those old puzzles where one tile is missing, and you have to move the tiles around to reveal the final image.

I turned this into a challenge: Can I go from the idea to App Store within a day? The first version should be as simple as possible. The answer is: Definitely.

First version

It took me 12 hours to go from creating the project in XCode to getting the app accepted by Apple. The main issue was not the app itself. It was rather to generate the images in a fast manner. In the end, I managed to generate a few decent images and published the first version with 34 images like these:

This is the final version. It’s very simple, in the sense that you see all the puzzles, select one and try to solve it. For now, I start counting the number of moves.

Major bug

After publishing the project, I tried to solve one of them. In hindsight, I should have done it during development. During development, I was focused on “shipping to the App Store” as fast as possible.

To generate the puzzles, what I did was to take all tiles and randomly shuffle them and mark the first tile as blank. Well it turned out, that in that case half of the puzzles are impossible to solve.

Luckily, there is a simple algorithm to check if any randomly generate puzzle is solvable. I spent another few hours to fix that bug and voilà, the new change got accepted today (02.10.2023) by Apple. 🥳

What is coming?

Don’t expect a lot! This was a learning project for me to improve my SwiftUI and AI skills. Depending on the interest of the app, I will decide to add more features.

From month on, I will go back to focus on the meal planner app and another project I have been working on (maybe you have already noticed?).

Thank you for reading this issue!

Thank you for reading this month’s newsletter. I wrote this issue a few days ago (so this one is exceptionally auto-scheduled).

The main focus of August was to close unfinished projects and reduce the number of projects I have to focus on in the future.

Too many projects

I like to start new projects. I like to bring them to a stand which can be used by anyone. Most of the projects are doing fine without me touching the code base. Most of them cost me barely any money to host or run. Some projects even have paying customers. For some, I was forced to put advertisements. For the others, I only touch them if there are major bugs.

Creating new projects is fun, but not a good long-term strategy if I want to be able to live off them. Therefore, I think it’s a good idea to focus on one or two projects in the long run.

This month, I started to clean up some projects. Most of them are tiny changes, but I am preparing to focus on a single project in the future. Here are all the small changes I made this month.

Changed ArticleToImage.ai slightly

ArticleToImage started as a simple idea: a tool which would give you the image if you give it the text. I started it as an experiment, but never completed the project. While going through all of my project, I thought why not make tiny improvements timeboxed (e.g. 1h) and leave it online.

In the long run, I will come back to this project when I feel it has potential (or if enough people are interested)

Changed Kiru.io design

I changed the design of my website slightly. Once again, I did a timeboxed session of two days to make small changes which would not cost me a lot of time and which would be visible to you.

I added a link to my newsletter issues and added images to all the blog articles. For some, I had to improve and add some fancy AI generated images :)

WhatDinner

I am preparing to finish the WhatDinner app (a meal planner for couples). There are a bunch of bugs, which I should fix before releasing it on the App Store.

I managed to add a search box to the recipe and make them editable. Well not much. I will continue on this project next month. My goal is to have it published by the end of the September.

ColorGuesser Traffic Spike

ColorGuesser was a simple project where you are given the name of a color and have to guess its RGB value. You get a score based on some closeness function. Since I put it online, the traffic has been growing:

This month, it got mentioned by UX DESIGN WEEKLY and has got a huge traffic spike ( I will let you know about the long-term effect next month):

That’s it for this month. Thank you again for reading this far.

This month, I went from learning SwiftUI to publishing a game to the App Store. More about that at the end of the issue. Let’s start with a few updates on other projects.

Reversle v2 improvements

Last month, I spent most of my time improving Reversle with additional features and bug fixes. The main change was the ability to play archived games. I limited them to the last 30 puzzles, with the option to buy the PRO option to remove them.

Until now, nobody paid for the PRO option. Ads seems the only way to keep the website online. My initial plan was to add the archive mode to MathleGame as well. But looking at the usage statistics of the new feature, I don’t think the development cost was worth it in terms of effective usage. Around 1% of the users used the archived puzzle feature. For now, I will put the plans on hold to improve MathleGame.

I don’t expect to make money with Reversle in the long run, therefore for the time being I use it as a billboard to promote some projects (some mine and some from Ed).

Learning SwiftUI

My background is backend and frontend development. Most of my projects are either done with React, Elixir, Java or Python (it doesn’t really matter which of them I use most). I would mostly not recommend doing Apps, but Ed hooked me on iOS Apps. After checking out SwiftUI, I decided to give it a try.

I got an iPhone, starting going through the SwiftUI Tutorial and porting WhatDinner to an App, it’s almost done.

At some point, I got bored with the tutorial, decided to work on a math game. This was the result:

You are given an equation and have to select the correct solution by given the first digit and then the second one. I gave it to some people to test, and the result was meh, but I got some great tips on how to improve and adapt. One of them was (by Ed) “what if you see some numbers and have to touch the correct one”. That got me thinking, and I prototyped the actual game I published on the App Store.

Math Multiplication Flash Game

I spent a total of~35 hours from creating the project to submitting to App Store. The submission process went surprisingly smooth, but creating proper screenshots and description to testing with different devices and figuring out what made sense (in terms of game logic, e.g. when to change the level, how to calculate the score) took the bulk of the 35 hours.

Here is the gameplay (sadly you don’t see me touching the solution)

The “game” is very simple: You are given an equation and a bunch of numbers. Your goal is to click the solution. More numbers appear if you are correct. In case you are wrong, a row disappears. After the second row, you have a time limit, if you don’t find the solution within the time, a row disappears as well. Once you hit a certain number of rows, you “restart” with a shorter timer.

I experimented with the number of rows and timer to find a sweet spot which worked for me. Just to keep me engaged enough and keep it challenging as well. The final app is not perfect, but I think it is a good basis to improve on. Click on the following link to download the app:

Let me know if you have any feedback on how to improve.

A few notes

Other links:

• I stopped posting original thoughts on Twitter (ähm, I mean X)

  • Why? I don’t know if it will be online in the future.

  • What then? I will write more on my website and write summaries on Twitter and LinkedIn

    • That way even if one of them disappears I can search through the content I wrote

• Checkout FlashDictation (German) if you are looking for an online dictation tool!

• Don’t forget: GeoTagSEO has a summer sale going on:

  • Get 50% off with discount code SUMMER2023

Thank you

Thank you very much for reading this issue of the newsletter. As usual, don’t hesitate to contact me if you have any questions or suggestions.

Thank you for signing up to this newsletter. This month is filled with tiny project updates and a new project in incubation phase. Maybe you have noticed it on my website.

PromptMonthly.com

Last month, I started another newsletter to send out inspirational AI prompts. After sending the first issue, I extended the website by adding some sample prompts and rewriting the FAQ section. It could still be improved, but for now, I am satisfied with its current state.

In terms of subscriber growth, we are at ~40. Granted, not many, but I also didn’t advertise it anywhere except on ProductHunt and Twitter.

For the ProductHunt launch, I experimented with creating promotional videos (using Remotion). Now I have a simple script that allows me to easily create videos like the following. My hope is to be able to re-use it for all future projects.

Reversle v2

Last month’s main focus was Reversle. Despite being an older project that I neglected, it still attracts around 500 daily users. A few months ago, I asked the visitors if they wanted an archive mode and how they wanted to support me. The answer was clear:

As you can see, a huge percentage is interested to play older puzzles. Apart from that, the huge majority is OK with having more ads. So this month I made those changes:

• Changed the favicons (the previous one was rushed)

• Added an Archive mode. Right now limited to the last 30 puzzles.

• Rewrote the instructions page

• Added FAQ section

• Redesigned the statistics page

• Added more ads (I am sorry, this is the only way right now to monetize the page and finance my development cost)

• Added a PRO option to see if anyone is willing to support me with a subscription

  • So far nobody is interested, I will keep it for a month and then decide what to do next

• Added a feedback form

  • I already got many good feedbacks and suggestions, but did not have the time to implement them yet

Once I am happy with the above changes and bug fixes, I will adapt them to Mathle, where even more people asked for the archived games.

Incubating ColorGuesser.com

A few years ago, I had this (potentially great?) idea: You are given a color name and a color picker, and your goal is to guess the corresponding color based solely on its name. I purchased the domain at the time, started prototyping, but never finished it.

While cleaning up some older projects, I rediscovered the code and thought, “Why not complete it and put it online?”. That is how ColorGuesser.com was born.

For now, I select 5 different color names and present them to you. You guess them and for each guess there is some score calculated. The basic idea was implemented within a few hours, but then I struggled with the actual scoring.

In the first version, I used the Euclidean distance between two RGB values, but that did not make any sense as a score. The calculated number did not match with my perception. Then I went down a rabbit hole, learning about color spaces and different ways of representing colors. People wrote research papers on that topic. For now, I settled on CIE94, but I am not very satisfied yet.

There are a few changes I wanted to make in the coming weeks:

• Write a blog article about color spaces (heck, I wasted too much time researching about it, at least the future me should profit from that)

• Cross-link between projects

• Add an unlimited mode

• Change the scoring system, try different color spaces and see if any of them make more sense

• Publish the project on ProductHunt and other pages

WhatDinner App

Many people have been requesting Reversle as an App. The current version is based on React, I could publish it as a React Native app, but lack the experience. A coworker motived me to look into iOS App development.

I started going through the official tutorial, but got bored and started porting another idea to iOS (aka learn by doing):

It’s a meal planner for couples who like to cook: You list all your recipes you ever cooked, and every day get an offer of three meals to choose from. This is how it looks like for now:

For now, I have a simple MVP. The idea is to have it working by the end of this month, and then I can actually start using the app daily.

A few other notes

• Wrote a single Today I Learned entry

  • TIL: Shortcut to forward standard output and error to another command

• GeoTagSEO has a summer sale going on:

  • Get 50% off with discount code SUMMER2023

• I reduced posting on Twitter. Used to be more active, but then I feel like Elon is destroying the platform. I’d rather spent more time publishing on my website

• To my surprise, SlashDreamer still growing slowly

  • I spent some time fixing bugs and replying to support mails

That’s it for so far. Thank you very much for reading this issue of the newsletter. 🙏 As always, feel free to contact me if you have any suggestions or questions.

Cheers,
Kiru.io

This is my first monthly newsletter on the projects I am working on. The main purpose is to keep you posted about any project updates. Apart from that, it is useful for me to sit down and reflect on the progress I made.

Starting PromptMonthly.com

There are two reasons for why I started an AI Newsletter. Firstly, I use MidJourney to build landing pages (some examples are: FlashDictation.com, WhatDinner.com and of course PromptMonthly.com) and one of the main issues was knowing how to write prompts. I thought it was easy, but then realized it’s more of a skill you develop. I thought it would be helpful to get some inspirations monthly.

Secondly: I went to a talk by Noemi about how she started FreshFonts.io and her story gave me the last push needed to try it out.

For now, I added a simple landing page, connected it with EmailOctupus and voilà. There are already 20 people who subscribed. Granted, not many, but I haven’t advertised it yet either. So, if you are reading this in your mail client and are also subscribed to PromptMonthly: Thank you so much for subscribing! 🙏

I kept the landing page on purpose “empty”. Once I published the first issue, I can add it to the main page and showcase how it should look like. The plan for next month would be to publish it more widely across the internet.

In case you are interested, this is how a newsletter looks like (I omitted the prompts):

Experimenting with Twitter Blue

I started taking Twitter more seriously and started tweeting at least every two days. The following charts show the effects:

There is a clear trend between my follower growth and my tweeting. My goal is to find out if there is any effect by having Twitter Blue, so I paid for the subscription for a month. I will continue tweeting every two days as well and see if there is any difference.

The following three tweets are the most viewed ones:

Updates on SlashDreamer

SlashDreamer is a project I did a few months ago. Initially as proof of concept to see if I can generate images directly via Notion. Published the project on Twitter and some people were interested, so I extend it to a full fledge service.

To my surprise, it still gets regular traffic with users subscribing to the service.

I fixed a few bugs this month (only relevant if you are a user):

• The quota was not correctly shown (it was not reset after each month)

• Upgrading to a different plan was not reflected in the code. That was fixed, now you get more quota once you upgrade.

• Performance improvement (due to Notion API limitation, I can’t generate images immediately, but I found a way to improve the performance nevertheless)

More ads on Reversle

I am not a fan of pages with a lot of ads, but I changed my mind about having ads on my websites. I am OK to put ads to cover my costs and development costs. Furthermore, I added an extra ad to cover any coming costs.

For Reverlse it seems a good pricing model (imho). I plan to spend more time in the coming days to work on Reversle. I have a backlog of requests from users, I haven’t found the time to do so.

Next month, I am trying to add the following features:

• Ability to play older puzzles (archive):

  • This is one of the most requested features!

• More statistics about play count and success rate

  • This would be useful for me to decide what features to work on in the future

• Add a proper help page with explanation video on how to play

• Add a resource section with helpful questions and answers

Personal website

One of my goals is to write more generally. I push myself to publish more online, even something tiny like a TIL (Today I Learned entry).

This month, I published the following articles:

• Eigengrau

• Unconventional Programming Opinion: You become an unofficial maintainer once you update your framework

• Extracting colors from images seems hard

The following chart shows the visitor count. I guess you can easily spot when I published the above articles:

Thank you

Thank you very much for reading this far. As usual, I am very happy for any feedback (via mail `feedback` at my website or @kiru_io on Twitter)