"You put water into a cup, it becomes the cup. You put water into a bottle, it becomes the bottle. Be water, my friend." — Bruce Lee
Water AI is the intelligent core of the Water ecosystem — a unified AI Supermodel designed to aggregate the world's best specialized AI capabilities into a single, accessible platform. Conceived by John Victor of the Stellarium Foundation, it is envisioned as a practical and accessible form of Artificial General Intelligence (AGI) and a gift to humanity.
Water AI serves as the brain of the ecosystem — a universal force-multiplier for human potential and a cornerstone technology for achieving the "Elevation to Eden." It understands complex requests, routes them to state-of-the-art specialized AI models (finance, law, engineering, creative arts, and more), and then acts — drafting contracts, generating 3D designs, composing strategies, coding software, and launching campaigns.
Water AI is not merely a tool — it is a platform that augments every human being with the combined intelligence of the world's best AI systems.
The AI revolution is here, yet its power remains fragmented. Users must navigate a complex ecosystem of specialized tools — one for writing, another for coding, yet another for research, and still more for creative work. General-purpose AI models lack deep expertise in every domain, while specialized models remain siloed and inaccessible to most people.
This fragmentation creates barriers:
- Knowledge workers waste time switching between tools and manually synthesizing outputs
- Professionals in law, finance, and healthcare lack AI that truly understands their domain
- Creatives cannot seamlessly move from ideation to execution within a single workflow
- Students and researchers struggle to access the best AI capabilities for their specific needs
- The general public is locked out of advanced AI by complexity and cost
The result: humanity's potential remains untapped, limited by the tools available rather than the ambition of the user.
Water AI solves this by acting as the Master Orchestrator. It provides a single, fluid interface that:
- Understands complex, multi-domain intent through natural language
- Decomposes tasks into actionable sub-tasks
- Routes each sub-task to the best-in-class specialized AI model
- Executes actual digital labor — creating documents, writing code, browsing the web, manipulating data
- Synthesizes results into coherent, high-quality outputs
- Iterates based on user feedback, maintaining full context throughout
Water AI intelligently routes requests to specialized models across finance, law, engineering, creative arts, and more — then acts on them, performing real work rather than just generating text.
| Differentiator | Description |
|---|---|
| Best-of-Kind Specialization | Dynamically leverages a curated ecosystem of the world's leading specialized AIs. Rather than one model trying to do everything, Water AI routes each task to the model that does it best — creating a dynamic ecosystem that evolves as new models emerge. |
| True Action & Labor Performing | Goes beyond text generation — creates, builds, and executes across digital tasks including document creation, code generation, web interaction, data manipulation, and creative media generation. |
| Open, Accessible & Potentially Client-Run | Open-source core components with the ability to run on client devices via WebAssembly for data sovereignty and privacy. Designed to be accessible to everyone, not just enterprises. |
| Intelligent Orchestration | Sophisticated Go-based AI core that plans multi-step workflows, selects the right tools and models, seeks clarification when needed, and maintains context across complex interactions. |
| Formless & Persistent | Runs as a background daemon on your OS, accessible via a native desktop GUI, web interface, or remote bridges — always ready, always adapting. |
- Professionals — Legal, Finance, Healthcare, Engineering, Business strategists who need domain-expert AI
- Creatives — Designers, Writers, Artists, Developers who need seamless ideation-to-execution workflows
- Researchers & Academics — Deep research, fact-checking, data analysis, and knowledge synthesis
- Students & Lifelong Learners — Learning augmentation, tutoring, and knowledge exploration
- General Knowledge Workers & Individuals — Everyday productivity, task automation, and personal AI assistance
- Chat interface supporting text, file uploads, and voice input
- Downloadable native desktop client (Windows, macOS, Linux) built with Fyne
- Web access via embedded frontend
- Multi-format output with user control and iteration
- Browser, code, and terminal panels for real-time visualization of AI actions
- Intent understanding and task decomposition
- Multi-model routing to specialized AIs
- Context management with token-aware windowing and conversation summarization
- Response synthesis across multiple model outputs
- Sequential thinking and planning modules
- Quality assurance via reviewer agent
- Slash commands for quick actions (
/help,/compact)
- Curated models from Hugging Face, open-source, and commercial sources
- Multi-provider LLM support: OpenAI, Anthropic (Claude), Google Gemini
- Domain-specific routing for finance, law, engineering, creative arts, and more
- Chain-of-thought reasoning support (o1/o3 models, Anthropic thinking tokens)
- Configurable model parameters (temperature, retries, token limits)
- Document Creation — Drafting contracts, reports, presentations
- Code Generation & Execution — Write, run, and debug software in sandboxed environments (Docker, E2B, local)
- Web Interaction — Browser automation via Playwright for research, data gathering, and web tasks
- Data Manipulation — Processing, analysis, and visualization
- Creative Generation — Image generation, audio transcription, media processing
- Client-side execution (WebAssembly/Python) and cloud-side execution paths
- Web search: Tavily, Jina, SerpAPI, DuckDuckGo
- Web scraping: Firecrawl integration
- Third-party APIs: Vercel, NeonDB, cloud storage
- Cloud platforms: Google Cloud Platform, Azure
- Planned: Email, social media, calendar, CRM, project management, and web service connectors
- Multi-format output rendering (text, code, images, files)
- Iterative refinement through conversation
- File management with workspace-per-session isolation
- Export and download capabilities
Water AI Architecture
┌─────────────────────────────────────────────────────────┐
│ CLIENT LAYER │
│ ┌──────────┐ ┌──────────────┐ ┌───────────────────┐ │
│ │ Fyne GUI │ │ Web Frontend │ │ WebAssembly (WASM) │ │
│ └────┬─────┘ └──────┬───────┘ └────────┬──────────┘ │
│ └───────────┬────┘ │ │
│ WebSocket / HTTP │ │
└──────────────────┬─────────────────────────┘ │
│ │
┌──────────────────▼───────────────────────────────────────┐
│ SERVER LAYER │
│ ┌─────────────────────────────────────────────────────┐ │
│ │ Gateway (Gin HTTP/WS Server) │ │
│ │ • Session Management • File Upload • Health API │ │
│ └────────────────────┬────────────────────────────────┘ │
│ │ │
│ ┌────────────────────▼────────────────────────────────┐ │
│ │ Orchestration Engine (Agents) │ │
│ │ • Prompt Builder • Context Manager • Reviewer │ │
│ │ • Task Decomposition • Tool Selection │ │
│ └────────────────────┬────────────────────────────────┘ │
│ │ │
│ ┌────────────────────▼────────────────────────────────┐ │
│ │ Tool & Action Layer │ │
│ │ • Browser • Terminal • File Editor • Search │ │
│ │ • Media • Code Exec • Web Scraping │ │
│ └────────────────────┬────────────────────────────────┘ │
│ │ │
│ ┌────────────────────▼────────────────────────────────┐ │
│ │ LLM Provider Layer │ │
│ │ • OpenAI • Anthropic • Google Gemini │ │
│ └─────────────────────────────────────────────────────┘ │
│ │
│ ┌─────────────────────────────────────────────────────┐ │
│ │ Infrastructure: SQLite/GORM • Sandbox (Docker/E2B) │ │
│ │ Config • Logging • Migrations • Process Manager │ │
│ └─────────────────────────────────────────────────────┘ │
└──────────────────────────────────────────────────────────┘
Client-side: Native Fyne desktop GUI + WebAssembly runtime for local execution
Server-side: Go-based orchestration engine with multi-provider LLM routing
Privacy by Design: Minimal data retention with client-side execution options. Envisioned to run high-performance actions locally on client devices to ensure data sovereignty.
Water/
├── cmd/water/ # Main entry point (daemon + Fyne GUI launcher)
├── server/ # HTTP/WebSocket server (Gin-based)
├── agents/ # AI agent abstractions (Base, Reviewer, FunctionCall)
├── browser/ # Headless browser automation (Playwright)
├── core/ # Core utilities: logging, config, event system, storage
├── db/ # Database layer (SQLite via GORM)
├── llm/ # LLM clients (Anthropic, Gemini, OpenAI)
│ └── context_manager/ # Token counting and context window management
├── migrations/ # Database migrations (Goose)
├── process/ # Process/session management & gateway
├── prompts/ # System prompt builder with domain-specific rules
├── sandbox/ # Sandboxed execution (Docker, E2B, Local)
├── tools/ # Tool implementations (browser, terminal, search, media, etc.)
├── ui/ # Fyne desktop GUI (chat, panels, settings, theme)
├── utils/ # Shared utilities (file manager, terminal manager)
├── resources/ # Logo and static assets
├── plans/ # Architecture and planning documents
├── .github/ # CI/CD workflows (GitHub Actions)
├── Makefile # Unified build system
└── ROADMAP.md # Detailed feature roadmap
| Revenue Stream | Description |
|---|---|
| Community Donations | Voluntary donations from individuals and organizations who benefit from Water AI |
| Enterprise Licensing | Volume licensing and tailored solutions for businesses requiring dedicated support and SLAs |
| Usage-Based Pricing | Fair pricing based on inference and cloud compute consumption for heavy users |
| Visionary | John Victor |
| Organization | Stellarium Foundation |
| Mission | Leverage technology for global prosperity and human advancement |
| Goal | The Elevation to Eden |
John Victor, founder of the Stellarium Foundation, conceived Water AI as the technological cornerstone of a broader mission to uplift humanity. The Foundation's work spans multiple initiatives, with Water AI serving as the intelligent core that ties them together.
Funding Ask: $600,000 for core platform development, foundational model integration, and key engineering talent.
Funds will be allocated toward:
- Core Orchestration Engine — Advanced task decomposition, multi-model routing, and planning capabilities
- Foundational Model Integrations — Connecting 500+ specialized AI models across every domain
- Key Engineering Talent — Hiring world-class Go, AI/ML, and systems engineers
- Infrastructure — Cloud compute for development, testing, and initial deployment
- Go 1.24+
- GCC / Clang (CGO is required for the Fyne GUI)
- Linux:
libgl1-mesa-dev,libxcursor-dev,libxrandr-dev,libxinerama-dev,libxi-dev,libxxf86vm-dev - macOS: Xcode Command Line Tools
- Windows: MinGW-w64
- Linux:
git clone https://github.com/StellariumFoundation/Water.git
cd Water
# Build the Go binary (no frontend/Node.js required)
make build-dev
# Run the server in headless mode
./bin/water-ai serverThe server starts on http://localhost:7777.
# Build the complete application with Fyne GUI
make build
# Run — launches gateway + native desktop GUI
./bin/Water
# Or run headless server only
./bin/Water server
# Check version
./bin/Water --versionmake test # Run all unit tests
make test-race # Run tests with race detection
make test-coverage # Generate HTML coverage reportmake release # Build optimized binaries for all platformsRelease binaries are output to the dist/ directory (Linux, macOS, Windows — amd64 + arm64).
See ROADMAP.md for the detailed, feature-level roadmap with implementation status.
- Core platform, multi-LLM orchestration, desktop GUI, tool framework
- Public API, 500+ specialized model integrations, MCP marketplace
- Community-driven marketplace, autonomous "Eden" workflows, full AGI capabilities
| Revenue Stream | Description |
|---|---|
| Community Support | Voluntary donations from individuals and organizations who believe in the mission |
| Enterprise Solutions | Volume licensing and tailored services for businesses requiring dedicated support |
| Usage-Based Options | Fair pricing based on inference and cloud compute consumption |
Funding Ask: $600,000 for core platform development, foundational model integration, and key engineering talent.
Funds will be allocated toward:
- Core orchestration engine development — Building the intelligent routing and task decomposition system
- Foundational specialized model integrations — Connecting to the best AI models across every domain
- Key engineering talent acquisition — Hiring world-class Go, AI/ML, and systems engineers
- Infrastructure and cloud compute — Development, testing, and production environments
Water AI's core components are open-source under the Apache 2.0 License. This strategy ensures:
- Transparency — Full visibility into how the AI operates
- Trust — Community-auditable codebase
- Community Contribution — Developers worldwide can contribute and extend
- Rapid Adoption — No barriers to entry for individuals and organizations
| Visionary | John Victor — Founder of the Stellarium Foundation, architect of the Water AI vision |
| Organization | Stellarium Foundation |
| Mission | Leverage technology for global prosperity and human advancement |
| Goal | The Elevation to Eden — using AI to unlock humanity's full potential |
John Victor conceived Water AI not as a commercial product, but as a gift to humanity — a platform that democratizes access to the world's most powerful AI capabilities, ensuring that every person on Earth can benefit from the AI revolution.
Water AI is an open-source project fostered by the Stellarium Foundation. We welcome developers who share the vision of human augmentation.
- Fork the repo
- Create your feature branch (
git checkout -b feature/AmazingAction) - Commit your changes
- Push to the branch
- Open a Pull Request
This project is licensed under the Apache 2.0 License — see the LICENSE file for details.
"Be water. Flow into the future."
