Artificial General Intelligence project (AGI)

AI KING is a comprehensive generative general intelligence platform powered by Large Language Models (LLM). The system features trained AI agents coordinated by an intelligent orchestrator, supported by a rich ecosystem of helpers and tools.

• Hovedidé: AI King er et sammensatt GGI-system, ikke én enkelt modell. Det bygges av flere spesialiserte deler som samarbeider.
• Orkestrering (Sovereign Layer): Bruk en deterministisk tilstandsmaskin med LangGraph og database-checkpointing for å kunne «time travel» ved feil. Kjør en hybrid arkitektur der LangGraph styrer, mens Microsoft AutoGen kan kalles inne i noder for kreativ idéutvikling.
• GGI-kjerne: Mixture-of-Agents (flere modeller parallelt) med en aggregator som syr sammen svarene. Korte læringssløyfer med Kritiker/Refleksjon, og meta-læring som automatisk oppdaterer agent-prompter når gjentatte feil oppdages.
• Hofstaten og verktøy: Supervisor-arkitektur med spesialiserte agenter (Researcher, Coder, Reviewer). Human-in-the-loop «breakpoints» før risikable handlinger, MCP som standard grensesnitt for verktøy, sikker kodekjøring i ekstern sandkasse (f.eks. E2B), og visionsstyrt nettlesing for å navigere komplekse websider.
• Hukommelse: Korttidsminne via checkpointing i LangGraph og langtidsminne via GraphRAG/Knowledge Graph for relasjonsforankret kontekst.
• Infrastruktur: Kubernetes for produksjon, AWS Fargate for langvarige agentløp, og skyvalg mellom Azure AI Foundry (Microsoft-integrasjon) eller AWS Bedrock/IaaS for maksimal kontroll.

• 🤖 Multiple Trained AI Agents: Specialized agents for domain-specific tasks
• 🎯 AI Orchestrator: Intelligent coordination and task management
• 🛠️ Tools Ecosystem: Comprehensive toolkit for AI operations
• 🔧 Helper Systems: Supporting infrastructure for robust operations
• 🧠 Generative Intelligence: Advanced LLM-powered cognitive capabilities
• 🏢 Enterprise-Ready: Scalable, secure, and production-grade

The foundation of AI KING, providing natural language understanding, generation, reasoning, and cognitive processing capabilities.

Central coordination system that manages agent lifecycle, routes tasks, allocates resources, and ensures optimal multi-agent collaboration.

Specialized AI workers trained for specific domains and tasks, capable of autonomous execution and collaborative problem-solving.

Extensive ecosystem of supporting systems and integration tools that enable agents to interact with external services and perform complex operations.

• QUICK_START.md - step-by-step local run guide (smoke tests, orchestrator, UI)
• TRADEMARK.md - Trademark information and system description
• NICE_CLASSIFICATIONS.md - Detailed Nice Classification analysis for trademark filing
• ARCHITECTURE.md - Comprehensive system architecture documentation
• ROADMAP.md - Implementation phases derived from Technical Blueprint
• Wiki - Repo-backed wiki with quick-starts, orchestrator guidance, and UI/server notes

AI KING is being registered under the following Nice Classifications:

• Class 9: Downloadable software, applications, and development tools
• Class 35: Business management and automation services powered by AI
• Class 42: Software as a Service (SaaS) and technology services

For detailed classification descriptions, see NICE_CLASSIFICATIONS.md.

• Business Process Automation
• Enterprise Knowledge Management
• Software Development Assistance
• Customer Service Automation
• Data Analysis and Insights
• Content Creation and Generation

• Cloud (SaaS): Fully managed service with API access
• On-Premises: Downloadable software for local deployment
• Hybrid: Flexible deployment across cloud and on-premises
• Edge: Lightweight deployment for real-time applications

Start with QUICK_START.md for a concise setup, smoke test, and UI run. The wiki remains available for a synced copy plus orchestrator tips. Installation, configuration, and usage documentation will expand as the platform develops.

Copyright © 2024 AI KING. All rights reserved.

For trademark inquiries, licensing, or general information, please contact the repository owner.

An initial deterministic orchestrator scaffold is available in the ai_king package. It includes typed state, checkpointing, and human-in-the-loop breakpoints as outlined in the architecture documents.

Run the smoke tests with:

python -m unittest

• A lightweight cyberpunk-themed HTML5 frontend lives in web/. Open web/index.html directly or serve it via the settings server.
• The settings server is a small C++ binary that serves the settings page and JSON APIs for prompts/templates.
• Build and run:

g++ -std=c++17 -O2 -o settings_server server/settings_server.cpp
./settings_server # defaults to port 8088

• Endpoints:
  • GET /api/prompts and POST /api/prompts read/write data/prompts.json
  • GET /api/settings and POST /api/settings manage backend settings
  • GET /settings serves the backend settings UI