A Next-Generation Agentic AI Simulation Platform
Powered by Cortana Research & Development Center
Descendants is an advanced 3D simulation platform designed to bridge the gap between human intelligence and artificial agents. Built on a "Modular Monolith" architecture, it combines high-fidelity 3D rendering with autonomous agent logic, enabling complex interactions in a persistent virtual world.
This project represents the foundation of CortanaOS—a future operating system for spatial computing and agentic collaboration.
Demo.2.mp4
Screen.Recording.2025-12-06.at.2.15.19.AM.mov
Our roadmap is driven by the convergence of Spatial Intelligence and Generative AI.
- Application Layer: User interfaces, Next-gen UI/UX.
- Communication Layer: Multi-Protocol Support (MCP), A2A (Agent-to-Agent) messaging.
- Intelligence Layer: Agentic AI, Large Model integration (500+ models), Adaptive Context.
- Spatial Layer: World Engine, Custom 3D Pipeline, Physics & Environment.
- Encryption Layer: Secure data handling for "Dual User" (Human & Simulant) privacy.
- Hardware Abstraction Layer: Cross-platform compatibility (Web, VR/AR).
- RealityLink: Seamless bridging of virtual and physical contexts.
- Simulated Vision: Agents possess depth perception, temporal context, and simulated "senses".
- Time-Travel: Advanced state management allowing temporal navigation of simulation states.
- Agentic Collaboration: Simulants (AI) and Humans working together in shared spaces.
The current codebase establishes the Core/Spatial and Intelligence layers:
- Rendering:
React Three Fiber(R3F) &Three.jsfor high-performance WebGL graphics. - Physics & Environment:
- Dynamic Time System (Day/Night cycles) with accurate Sun/Moon positioning.
- Procedural Terrain generation using FBM noise & Instanced foliage.
- Atmospheric Zone Controller for smooth Fog/Lighting transitions between biomes.
- Collision detection with Raycasting and Height-field physics.
- Brain: Integrated with Google Gemini 2.5 Flash for high-level decision making.
- Perception: Agents "see" nearby entities (Players, Agents, Obstacles) and their status.
- Reasoning: Agents autonomously decide to Follow, Wander, Chat, or Wait based on visual context.
- Motor Control: Powered by Yuka AI for autonomous steering behaviors.
Seek,Wander,Separation,ObstacleAvoidance.- Smooth navigation and separation logic for crowd dynamics.
- Social:
- Agent-to-Agent interactions (Greeting, Chatting states).
- Agent-to-Player interactions (Look-at, Waving, Following).
- Framework: Next.js 16 (React 19)
- Language: TypeScript
- State Management: Zustand
- AI/ML: Google GenAI SDK (Gemini), Yuka GameAI
- 3D: React Three Fiber, Drei
- Node.js 18+
- pnpm (recommended) or npm
- Google Gemini API Key
-
Clone the repository
git clone https://github.com/Cortana-Devs/Descendants.git cd Descendants -
Install dependencies
pnpm install
-
Environment Setup Create a
.env.localfile in the root directory:NEXT_PUBLIC_GEMINI_API_KEY=your_api_key_here
-
Run Development Server
pnpm dev
Open http://localhost:3000 to enter the simulation.
| Phase | Status | Feature |
|---|---|---|
| Phase 1: Foundation | ✅ | Core 3D Engine, Procedural Terrain, Day/Night Cycle |
| Phase 2: Intelligence | ✅ | Gemini 2.5 Integration, Yuka Steering, Basic Agent Behaviors |
| Phase 3: Social | 🚧 | Multi-Agent Chat, Group Dynamics, "Simulacra" Personality Engine |
| Phase 4: Expansion | ⏳ | Infinite Terrain, MCP Integration, RealityLink Prototype |
| Phase 5: CortanaOS | 🔮 | Full Operating System Logic, Async Priority Queues, Deep Memory |
We welcome contributions from industry experts and engineers.
- Fork the repository.
- Create your feature branch (
git checkout -b feature/amazing-feature). - Commit your changes (
git commit -m 'Add some amazing feature'). - Push to the branch (
git push origin feature/amazing-feature). - Open a Pull Request.
Distributed under the MIT License. See LICENSE for more information.
Designed and engineered by the Cortana Research & Development Center.