How can Gemini help generate dynamic NPC dialogue for life simulation games like Tomodachi Life?

Hi everyone,

I’m experimenting with AI-generated dialogue for a life simulation game concept inspired by Tomodachi Life.

I’m interested in using the Gemini API to create NPC conversations that:

  • Remember previous interactions.
  • React differently based on personality traits.
  • Generate natural and varied dialogue.
  • Keep responses family-friendly and consistent.

Has anyone built a similar system with Gemini?

I’d appreciate advice on:

  • Prompt engineering strategies.
  • Context and memory management.
  • Structured JSON outputs for game events.
  • Reducing token usage while maintaining conversation quality.

If you have sample prompts or best practices, I’d love to learn from your experience.

Thanks!