We needed to rebuild our own apps

We started Glowbom to try to turn everybody in the world into software engineers. But back then, we did not know exactly what kind of product to build to achieve that goal.

My journey with software started when I was 12, when I built my first computer game. It was a soccer manager for a ZX Spectrum-like computer, and many of my friends played it and liked it. I also found it very interesting to build mini-programs for personal use using C, C++, Assembly, and BASIC.

My mom is a librarian, and she also worked at a book market back then, so she gave me access to technical literature. Reading very technical books about languages like C++ as a kid made me curious about new technology in general. I believe that is how I became interested in building software for mobile phones, even before the iPhone, for Nokia phones and the J2ME platform.

The big shift happened when the App Store and Android Market launched in 2008. That is when I became hyperfocused on launching different ideas for iPhone and Android. I built apps as an indie developer and later as a startup engineer in Silicon Valley.

Most of the things I shipped as an indie failed miserably as products, but the very few that succeeded changed my life. I did not know anything about marketing, but eventually I got lucky with a few games like Glow Puzzle and Office Story. At one point, Glow Puzzle was making around $20,000 a month from different sources, while my living expenses were only about $500 a month. That was huge for me. I thought it would be great if more people were able to build software so they could change their lives for the better too.

The idea behind Glowbom actually started with my sister. She wanted to build apps too because she saw me building apps and becoming successful to a certain extent. She did not know how to code, so she asked me to create something like an engine for her.

I built a customizable codebase for her in Unity because I was using Unity for games. She released a number of apps this way. They were not hugely successful, but one app was downloaded more than 50,000 times. I think that is where the idea behind the company started.

Over the years, we built more than 50 different apps and games. The most successful were Glow Puzzle (#2 free game in the US App Store, millions of downloads), Office Story (featured worldwide on Google Play, millions of downloads), and Citizenship Test (#1 paid app in the Education category on the US App Store due to TikTok success, 300k plus followers and millions of views).

Many of our software projects were either never released or failed and were taken off the market. As a result, we accumulated a large number of projects built with outdated frameworks. Some were even built for systems where it no longer makes sense to release software, such as Symbian and Windows Phone.

Some of our apps had a lot of downloads, but success is relative. Today, many of them probably have very few active users and generate only a little revenue every month. Is that success? Probably not.

But the opportunity here is gigantic. We can use our old codebases as references and rebuild these projects with new AI tools and systems. We are building the tooling for ourselves because we are still running an indie software development company. If the system works for our projects, other people can use it for their projects too.

Today’s AI tools are amazing and have made so much progress over the past few years, and even in recent months. However, they are still a little too complicated for the general public.

We wanted to find a way for people to build software more easily using the AI systems already available on the market. It is hard to absorb all the ideas contained in an old software project at once. We wanted to create one standard that agents and humans could easily collaborate on, so they could understand why and how we are rebuilding each project with AI. We call it the Project Book, and it is based on the Glowbom standard.

A Project Book can contain vector drawings represented as coordinates, images such as PNG or JPEG files, Markdown files, and sometimes JSON files. Together, these materials feel like a book with certain rules. The Project Book should be understandable to humans while also making it clear to AI agents how to work on a project, why we are building it, why certain decisions are important, and why the project itself matters.

A regular person can create a narrative and brainstorm an idea with AI. AI agent teams can then put that idea to work. They can not only create the software but also help build a company and business around it. The person develops the project almost like writing a book. They create the narrative, sketches, plans, and text. Another analogy would be the engineering documentation for a house. The Project Book contains the sketches, plans, and written explanations. The AI agents working through different systems are like the builders who use those plans to build the house.

I think this is a golden age for indie developers and very small teams. One person can connect multiple AI employees working through Claude Code, Codex/ChatGPT, OpenCode, and other systems. Those employees can communicate through a project channel in Buzz.

I think I am on the spectrum, and I believe I may also have ADHD. This format helps me organize my ideas, stay focused, and actually finish things. I think it could also be useful for other neurodivergent entrepreneurs. You might ask your AI employees what they think about a feature, brainstorm how it should work, and decide how to build it. Glowbom summarizes those discussions and stores them in the Project Book. When people and agents return to the project, they can understand what was previously discussed, what decisions were made, and why.

This could make small organizations much more scalable. Maybe we will see one-person or five-person billion-dollar companies. But success does not need to mean building a billion-dollar company.

If one person builds something that makes $3,000 or $5,000 a month, they may be able to live from that income in many parts of Europe or elsewhere in the world, outside the most expensive cities. That is also success. Our mission is to make this more accessible to the general public.

Our first major experiment is rebuilding Office Story. We ported the old Unity project to Godot. The engine we developed while rebuilding it is also being used for Glowbom Live, a multiplayer AI simulation where people can see and communicate with different AI employees working on their projects. One of my earlier ideas was to let people build apps through a game, and in 2026, current AI systems finally make that idea possible.

Over the years, I had several ideas for a sequel or other versions of Office Story, and I think now is the perfect time to build them. This type of game needs an interesting story, compelling graphics, 3D models, and marketing materials to succeed. Specialized AI tools can now help with all of those things. Glowbom brings them together through one project structure. Office Story is our first experiment, and multiple AI employees are working on it.

Another big part of our philosophy is freedom. You should be able to take full control of your codebase and your creation. You should be able to keep it locally and take the source code and assets to whatever tools you want to use. We simply provide you with guidance.

The open-source version of Glowbom is already available. We will gradually announce other parts of the system as they become ready to use. For now, we are using Glowbom to rebuild our own projects and to see what the product needs to remember, what it should make easier, and where it still falls short. We can use our old codebases to rebuild those products, and other people can use the same system for their own ideas.

Hopefully, that becomes an interesting business by itself.