I'm really impressed by how effective GPT 4.1 is for Unity development when used alongside Visual Studio Code. I feel like this combination is often overlooked! The secret seems to be using Gemini 2.5 Pro Experimental with my entire code base, including game rules and narratives. It helps me generate instructions for changes—whether they're small tweaks or major features—which I then input into the 4.1 Agent in VS Code. This approach has truly transformed my workflow; almost every change I make is implemented successfully on the first try. Has anyone else experienced similar boosts in productivity when working with Unity?
4 Answers
That's really interesting! So you ask Gemini for step-by-step instructions and then GPT does the actual coding? What a clever setup!
Great post! Thanks for sharing your experience.
To share your whole codebase, how exactly do you do it? Are you zipping the entire project or using a tool like Co-Pilot? Thanks for the insights!
I’ve found success using the RepoMix tool from GitHub! It allows for easy integration, which I highly recommend.
I remember doing something similar a while back. If you search for scripts that can merge folders into a single file, it fits snugly into the context for Gemini 2.5. After that, I just branch off my prompts and ask questions related to each specific task.
Could you give more details on how you access and integrate your codebase with AI? This sounds intriguing!
I created a single file that combines all my scripts, game rules, and other relevant information. Then I ask Gemini questions about how to implement changes. The instructions it provides go directly to the 4.1 agent, and I’ve had minimal errors—unlike previous experiences with coding AI.
Exactly! While 4.1 isn’t perfect at suggesting big changes, it really shines when following instructions. This combo of guidance and precise coding has been a game changer for me!