Konbini Days

Inspiration

Thinking Simulation and Management realm: My favorite games of all-time is the Age of Empires franchise... So I thought maybe make a store sim where you can place the shelves like buildings in a RTS on a grid like map and have AI units or customers shop. So that was the initial concept and this is what it turned into...

What it does

The player controls a single owner who must choose, moment to moment, between restocking and serving. Standing at the register keeps the checkout moving; walking off to refill a shelf stops it. Across one escalating day after another, the goal is to serve as many customers as possible, meeting the store reputation rating requirement and money balance for that day, before closing while spending cash on upgrades. Feedback is immediate: stock-level fills and empty-badge rings on every fixture, visible customer patience states, cash ticking up per sale, and reputation penalties for messes or abandoned items. It's satisfying to repeat because the tension is spatial and numeric.

How we built it

Built by just talking to "Sir Claude" Fully built using Claude, Sonnet 5 Extra Used World Desktop Editor to make the stand-in 3d models, then uploaded to Claude project

Challenges we ran into

I think the bottleneck most of the time is not credits or development time at this prototype stage, it is more of the questions and decisions of what should I add next, or how should the game run, or is there a better way to go about it. Developing with AI make it seems like everything is a possibility in terms of being realized, but now the question is what do you intend the outcome to be, how should it play, because you change any part of your game or all of it in an instance with a few sentences.

Accomplishments that we're proud of

All the structural details to the game and different features that were added, would have been completely undoable in the allocated time without AI. Adding the probability system for the customer AI's behavior, building AI's with AI is actually fun cause of how easy it is now (at a basic game level)

What we learned

AI is crazy for building games, building in natural language is here, however, fundamental understanding of game development, modeling, and design are still needed if you want to build with specific intention and build efficiently. AI is great for building complex probability systems and behavior trees in an instance.

What's next for Konbini Days

Build the details, develop a strong consistent visual style, add a ton of details to the behaviors, experiment with different game structures for the environment, build out the rest of the Konbini Days world, and ultimately build it for mobile in Meta Horizon with the new AI tools hopefully, it will be interesting to see how it will be to convert from a Claude build to Meta build.

Built With

• chatgpt
• claude
• worlddesktopeditor