100% AI Coded Game on Steam – Full Dev Log(web3dev1337.github.io)
web3dev1337.github.io
100% AI Coded Game on Steam – Full Dev Log
https://web3dev1337.github.io/epic-survivors-architecture/
4 comments
Thanks proc0!
Totally get where you are coming from, but there is no way I could have coded anywhere near as fast without AI. The game has no 'engine', it is built from the ground up. While there are a few 'god classes', and a little spaghetti, it's overall pretty decent. I can run many agents in parallel without merge conflicts due to the fact it's well separated and modular. It's closer to a game engine for Survivor Roguelikes than a game itself. The bottleneck was actually balance, sound, music, art etc.
The article goes into more details, but you will find that AI can easily handle it if you use the techniques I discussed :) Let me know if you got questions about that further, I will continue to do more coverage into it over time, as there are plenty of techniques that help out (many in the article, some not :D).
But still expect to put in plenty of work, it's not some 1 shot machine...although maybe Ralph Loops will change it soon...which I have experimented with!
Totally get where you are coming from, but there is no way I could have coded anywhere near as fast without AI. The game has no 'engine', it is built from the ground up. While there are a few 'god classes', and a little spaghetti, it's overall pretty decent. I can run many agents in parallel without merge conflicts due to the fact it's well separated and modular. It's closer to a game engine for Survivor Roguelikes than a game itself. The bottleneck was actually balance, sound, music, art etc.
The article goes into more details, but you will find that AI can easily handle it if you use the techniques I discussed :) Let me know if you got questions about that further, I will continue to do more coverage into it over time, as there are plenty of techniques that help out (many in the article, some not :D).
But still expect to put in plenty of work, it's not some 1 shot machine...although maybe Ralph Loops will change it soon...which I have experimented with!
Thanks for testing it out CatsOnHats
I agree the game is far from original or as polished as it needs to be, that is why it is a Demo. I've fixed many things since the initial launch, which was a marathon 14 week sprint of 1300 commits, and 8-16 hour days, which you can see in the article.
This is my first game I made, and I am proud of the time it was made in, and the fact that it proved that it is possible to use AI coding to make a real game, something people barely believe now, let alone in October 2024 when the majority of the engine was coded. The quality of 'games' going viral on twitter that are 'vibe coded' are an absolute joke, and closer to the 'slop' you mention.
I understand most gamers are anti-ai, but it is what it is, I accept it...we got 400 wish lists, so can't hurt :)
I am aiming to significantly improve it by the full launch, hope you will give it 1 more try before launch :)
I agree the game is far from original or as polished as it needs to be, that is why it is a Demo. I've fixed many things since the initial launch, which was a marathon 14 week sprint of 1300 commits, and 8-16 hour days, which you can see in the article.
This is my first game I made, and I am proud of the time it was made in, and the fact that it proved that it is possible to use AI coding to make a real game, something people barely believe now, let alone in October 2024 when the majority of the engine was coded. The quality of 'games' going viral on twitter that are 'vibe coded' are an absolute joke, and closer to the 'slop' you mention.
I understand most gamers are anti-ai, but it is what it is, I accept it...we got 400 wish lists, so can't hurt :)
I am aiming to significantly improve it by the full launch, hope you will give it 1 more try before launch :)
I tried the demo during next fest and found it frankly uninspired slop. There are a thousands of survivors like games, and this doesn't even remotely stand out.
Also advertising this when your customer base is gamers who are notoriously anti ai in gaming is a terrible decision.
Also advertising this when your customer base is gamers who are notoriously anti ai in gaming is a terrible decision.
One observation so far is seeing that there is real effort in configuring the agentic workflow, and getting that to work even semi-consistently, meaning that there is always tweaking and experimentation for the specific use case. This isn't trivial work, and arguably it's a tradeoff, since all that effort could be spend coding the game directly. With enough skills it would be about the same time except better quality to code it by hand (with of course AI assistance where needed, but it would be human written).
This just means that the real value so far with AI is in letting non-programmers do coding, but it doesn't come without the cost of real effort to set things up and less quality in the end.