| Criteria | Rank | Score* | Raw Score |
| Production Velocity | #2 | 4.000 | 4.000 |
| Theme Fidelity & Creativity | #5 | 4.154 | 4.154 |
| AI Direction & Adaptability | #6 | 3.923 | 3.923 |
| Overall Enjoyment | #7 | 3.769 | 3.769 |
Ranked from 13 ratings. Score is adjusted from raw score by the median number of ratings per game in the jam.
The Biggest Challenge
Keeping everything feel like one game. The art came from image generation in separate batches — characters, title scene, logo, fan, farmer animations, store art — and each batch drifted a little in style, scale and lighting. The breakthrough was treating the AI like a small team with a strict brief: every asset request came with exact canvas sizes, registration points and a palette, and every delivery was checked in the actual game before it was accepted. The same loop worked for design: two live playtesters said they didn’t understand the goal, and within an hour the title screen had a slogan, the how-to-play card had a goal line, and weather got explained in the HUD. AI also ran hundreds of simulated seasons to tune beetles, tension and rival weights, and caught a browser memory leak that would have crashed the tab after 30 seconds.
AI Tools Used
GPT-6 Astra — first playable version, artwork and animation assets, UI refinement, itch.io page graphics, build packaging. Claude (Anthropic) — game code and fixes after the first version, QA and browser playtesting, balance simulations, in-game UI, audio looping, screenshots/GIFs and store page text. Suno — music.
Areas Where AI Was Used
Code (first version by GPT-6 Astra, then fixes, features and balance by Claude in Godot / GDScript), QA and balance (simulated seasons), UI, images and animation (characters, title scene, logo, farmer and fan animations, store page art), music (Suno tracks, looped and mixed with AI help), store page text and media. Fonts are free human-made fonts (Kaiju Monster, Barlow, Gochi Hand).
This was super fun! As you can see in the screenshot, I somehow managed to grab 1st place!
My strategy was rotating water, fertilizer, fans, and rutabagas, then fine-tuning from there. Whenever micro-cracks appeared, I'd stop using GROW and repeat that cycle.
The artwork is fantastic—the pumpkin, characters, and overall world-building feel so cohesive and well-integrated. It's really well done!
If I had to suggest a few improvements:
Awesome game!
Congrats on first place! 🏆 Your strategy is exactly the loop I was aiming for - rotating resources and pausing GROW on micro-cracks is the intended rhythm.
On hazards: there is weather in the game, so if you didn’t spot it, that’s a readability problem on my end. Good to know.
And the camera suggestion is a great call - starting tight and pulling back until the pumpkin doesn’t fit the frame would sell the growth far better than extra stages. Thanks for the detailed feedback!
Leave a comment
Log in with itch.io to leave a comment.