This is a tailored draft for your hackathon submission, organizing the technical and narrative details of ZINDAGI into the specific categories you requested.InspirationThe inspiration for ZINDAGI (Life) came from a simple question: What if a game could truly remember you? Traditional RPGs rely on static branching paths that feel restrictive. We wanted to create a "living simulation" where the environment doesn't just react to your buttons, but understands your intent. We were inspired by Chaos Theory—the idea that a small conversation in "Turn 1" could lead to a massive geopolitical shift in "Turn 50." With Gemini 3 Pro, we finally had an AI "brain" powerful enough to act as a sentient Game Master.What it doesZINDAGI is an immersive Adventure Quest where an AI "Overseer" controls the chaos. Unlike scripted games, every player choice—from dialogue to combat—is processed to calculate real-time environmental and social consequences. The game tracks your "Chaos Level" and adapts the world’s difficulty, NPC attitudes, and narrative events based on your unique playstyle, ensuring no two playthroughs are ever the same.How we built itWe built ZINDAGI using a high-performance modern stack:Frontend: Next.js hosted on Vercel for a seamless, low-latency interface.Intelligence: Gemini 3 Pro API handles all narrative generation and logical reasoning.Chaos Logic: We implemented a "Consequence Formula" using LaTeX logic: $C = \int (f(p) \cdot \Delta w) dt$, where the AI calculates the integral of player behavior over time to shift the world state.Memory: We used Gemini’s long context window paired with a custom "Recursive Summarization" system to maintain a perfect long-term memory of player actions.Challenges we ran intoContext Management: Managing an "infinite" story required us to condense past events into "Core Memories" so the AI wouldn't lose track of the plot.The "Overseer" Persona: It was a challenge to prompt Gemini to be both a helpful narrator and a challenging antagonist without it feeling unfair or robotic.Latency: To keep the adventure immersive, we optimized API calls using streaming responses so players see the story unfold in real-time as the AI "thinks."Accomplishments that we're proud ofTrue Emergent Gameplay: We successfully moved away from "If/Else" coding to a system where the story is synthesized on the fly.Dynamic World State: Seeing the AI autonomously decide to burn down a village or start a revolution based on a player's minor lie was a huge "Eureka" moment.Vibe Coding: Leveraging Gemini 3's ability to understand intent allowed us to build complex game mechanics in record time.What we learnedWe learned that the future of gaming lies in Agency, not just graphics. Harnessing Gemini 3 Pro as a Logic Engine rather than just a chatbot showed us that AI can manage complex, ever-evolving ecosystems. We discovered that when you give an AI control over the "chaos," the resulting stories are often more human and unpredictable than any scripted plot.What's next for ZINDAGIWe plan to expand the "Overseer" into a multimodal entity, allowing it to generate procedural images of the chaos using Nano Banana. We also want to implement a multiplayer "Shared Chaos" mode where multiple players' ripples collide in the same simulation, creating a truly global AI-driven ecosystem.

Built With

  • 3
  • ai
  • ai-driven-simulation
  • allowing-the-overseer?s-narrative-to-unfold-in-real-time
  • and-maintaining-the-simulation's-"memory"-through-its-massive-1m+-context-window.-vercel-ai-sdk:-used-to-implement-streaming-responses
  • and-responsive-interface-that-adapts-to-the-shifting-"chaos-level"-of-the-game.-vercel:-our-deployment-and-hosting-platform
  • api
  • atmospheric
  • css
  • especially-when-handling-the-complex-json-structures-of-the-"world-state."-tailwind-css-&-shadcn/ui:-for-building-a-sleek
  • gemini
  • long-horizon-planning
  • lucide
  • next.js
  • pro
  • sdk
  • tailwind
  • typescript
  • utilizing-the-app-router-and-server-actions-to-handle-complex-narrative-state-transitions-and-provide-a-lightning-fast-user-experience.-gemini-3-pro-api:-the-"brain"-of-zindagi.-we-used-this-for-advanced-reasoning
  • vercel
  • we-leveraged-a-cutting-edge-full-stack-ecosystem:-next.js-15:-our-core-framework
  • which-significantly-improved-player-immersion.-typescript:-used-throughout-the-project-to-ensure-type-safety
Share this project:

Updates