I released a simple shooting game called Bullet Rain.

Reasons for Starting to Make the Game

I have loved games for a long time and wanted to try making one.

The field of gaming does not require facing the 'needs' that individual developers and startups often struggle with. There certainly are needs, and I believe that with the future development of AI, humanity will have more leisure time, which will strengthen those needs even further.

Instead, the concept of 'fun' exists in games. From a certain perspective, it may be much harder than 'needs.' However, I have a sense that 'fun' is easier to replicate, so I decided to give it a try.

The Journey to Release

In fact, I had been tinkering with Unity and other tools for several years, but due to my lack of ability or losing sight of what 'fun' is, I had hit a wall.

Since the emergence of generative AI, implementation has become easier, and I have been able to create materials again, which has boosted my motivation. However, I have often been overwhelmed by large-scale ideas, leading to setbacks. While there are no issues with static assets like 2D images as items, animating them is currently challenging. 3D graphics have become relatively easier to create, but animation is still tough.

However, the logic aspect of games is a domain where AI excels, making it considerably easier. In games that do not require material, I can delegate most of the work to AI.

So, I decided to create a game that meets the following conditions:

1. A game that does not require animated assets
2. A game that feels fun with simple logic
3. A game that can be released as small as possible

What I Wanted to Test This Time

1. The Gap Between My Perception of Fun and the General Public's Perception

There are various things expected from a game, but I believe the most important is 'fun,' or 'gameplay.' This is highly subjective, so I wanted to first confirm how much my perception of fun resonates with others.

If the difficulty is perpetually challenging, people might get bored, so I hypothesized that a game where you quickly game over if you let your guard down might be more motivating. I thought a difficulty level that could be sustained to a certain extent but gradually becomes tougher over time would be ideal.

After releasing it and posting it on X, it didn't particularly go viral, but it seems over 100 people played it collectively more than 1,000 times. I received feedback through comments and RP stating, 'I got hooked on it!' (I'm happy!)

I made a simple but subtly difficult game!
If you score over 3000 points, I think it's quite impressive!
Give it a try to kill some time🎮https://t.co/gxqFFRnQqa

— Kazuki Shibata｜microCMS (@shibe97) July 24, 2025

2. How Far AI Can Take Production Quality

Claude Code is an incredibly great tool, but as I tried to create larger games, I felt my understanding diminishing as the code base grew. If that’s the case, could I release a smaller game? I wanted to test that.

Especially concerning security. I created this as a browser game and since it's a single-page application, various communications can be viewed in the network tab of the developer tools. I believe this is a significant difference from native apps, and to prevent cheating in games, you generally need to increase server-side processing. However, doing so might hinder smooth operation and compromise the UX.

Thus, it was a challenge to see if I could maintain adequate security while considering these factors.

When using AI without guidance, it often included unnecessary components in the API responses, resulting in potential cheating. This is where human oversight is still necessary.

3. The Effectiveness of Virality

I believe virality is essential for the growth of individual-developed games. Hence, I created a score ranking feature where players can share their results if they achieve a good score or get a high rank.

Just got 3,800 in BULLET RAIN! Can you beat me? https://t.co/wivvRCjtem #BulletRain

— Kazuki Shibata｜microCMS (@shibe97) July 24, 2025

Since sharing on social media is crucial for creating virality, I also implemented a system where OG images are automatically generated based on scores. This part was quite complicated when run on Cloudflare Workers and challenging with Claude Code, so I handled it personally.

The sharing feature took the most time...but honestly, it hasn’t been used much at this point.

It reaffirmed that there are indeed some hurdles to sharing.

What I Want to Verify Moving Forward

1. Whether It Can Appeal Internationally

From the perspective of 'fun,' I felt it might have some potential, so I want to test if this appeals internationally. I also feel simpler games are less likely to have discrepancies from domestic ones.

I think advertising can make it easy to test this.

2. Whether It Can Be Monetized

User payment is a high barrier, and as mentioned regarding security in browser games, it’s quite challenging. Although I don’t want to, I think monetizing through ads will be one way forward.

This would make DAU and PV the KPIs.

Games that attract users through ads and monetize through ads are often seen in recent overseas games, but this seems to come with quite a few hurdles.

My personal plan is to first get a significant number of players through ads and then wait for some virality to emerge from that. If a cycle of virality continues, it will create a situation where there is a steady inflow without advertising, so I think that is my goal.

Once in that state, monetization through in-game ads will undoubtedly be successful.

So, I plan to continue working steadily on this.