The world's first self-evolving autonomous intelligence platform
While others build static workflows, intelligentOne evolves them autonomously.
intelligentOne represents a fundamental paradigm shift in how AI systems operate. This isn't another automation tool—it's the first platform that writes, tests, and deploys its own capabilities autonomously.
| Traditional Systems | intelligentOne |
|---|---|
| Static, hardcoded workflows | Self-evolving, autonomous workflows |
| Manual tool integration | Dynamic MCP-native tool discovery |
| Human-designed pipelines | LLM-generated & LLM-judged recipes |
| One-size-fits-all solutions | Context-adaptive intelligence |
| Siloed capabilities | Composable atomic tools |
| Degrades over time | Improves continuously |
intelligentOne operates through three revolutionary layers:
The creative core that generates workflow ideas using LLM reasoning:
- Analyzes objectives and context
- Generates multiple workflow hypotheses
- Combines atomic tools in novel ways
- Learns from successful patterns
The quality gate that validates workflows before deployment:
- Executes hypotheses in isolation
- LLM-as-Judge scoring (Completeness 30%, Relevance 40%, Actionability 30%)
- Automatic deployment threshold: 85+ score
- Zero risk to production systems
The memory that stores and evolves successful recipes:
- Persistent JSON storage of proven workflows
- Hot-reload capability for dynamic composition
- Success metrics and performance tracking
- Continuous improvement through execution feedback
┌─────────────────────────────────────────────────────────┐
│ User Query/Objective │
└────────────────────┬────────────────────────────────────┘
│
┌───────────▼──────────┐
│ Router Agent │ Intent Classification
└───────────┬──────────┘
│
┌───────────▼──────────┐
│ Blueprint Vault │ Check for existing
│ (Search) │ workflow match
└───────────┬──────────┘
│
Found? │ Not Found
┌───────────▼──────────┐
│ Execute Blueprint │
└──────────────────────┘
│ Not Found
┌───────────▼──────────────┐
│ Hypothesis Engine │ Generate new
│ (LLM generates ideas) │ workflow ideas
└───────────┬──────────────┘
│
┌───────────▼──────────────┐
│ Simulation Lab │ Test & Judge
│ (LLM-as-Judge) │ Score: 0-100
└───────────┬──────────────┘
│
Score >= 85?
┌───────────▼──────────────┐
│ Blueprint Vault │ Auto-deploy
│ (Store & Deploy) │ successful recipes
└──────────────────────────┘
# Clone the repository
git clone https://github.com/groupthinking/intelligentOne.git
cd intelligentOne
# Create virtual environment
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install dependencies
pip install -r requirements.txt
# Configure environment
cp .env.example .env
# Edit .env and add your API keys (OpenAI or Anthropic)Add to your Claude Desktop MCP configuration:
MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"intelligentone": {
"command": "python",
"args": ["/absolute/path/to/intelligentOne/intelligentone_server.py"],
"env": {
"OPENAI_API_KEY": "your-key-here",
"ANTHROPIC_API_KEY": "your-key-here"
}
}
}
}Restart Claude Desktop and you'll see intelligentOne tools available.
# In Claude Desktop, try:
"Monitor TechCrunch RSS for AI chip announcements and alert me about urgent ones"
# intelligentOne will:
# 1. Classify the intent (actionable)
# 2. Check for existing blueprints
# 3. Generate a new workflow hypothesis if needed
# 4. Test it in Simulation Lab
# 5. Auto-deploy if score >= 85
# 6. Execute the workflowQuery: "Monitor competitor product launches and calculate competitive delta"
intelligentOne creates:
- RSS monitoring workflow
- OGP metadata extraction
- Competitive analysis scoring
- Urgent alert system
Query: "Track AI chip industry news and identify breakthrough capabilities"
intelligentOne builds:
- Multi-source RSS aggregation
- Keyword-based filtering
- Capability extraction from articles
- Trend analysis and alerts
Query: "Find and summarize latest developments in quantum computing"
intelligentOne evolves:
- Information gathering workflow
- Content extraction and analysis
- Summary generation
- Knowledge base updates
| Component | Technology | Purpose |
|---|---|---|
| MCP Server | FastMCP | Native tool protocol integration |
| Hypothesis Engine | GPT-4o / Claude 3.5 | Workflow generation |
| Simulation Lab | LLM-as-Judge | Quality assurance |
| Blueprint Vault | JSON Storage | Recipe persistence |
| Web Scraping | httpx + BeautifulSoup | Content extraction |
| Feed Monitoring | feedparser | RSS/Atom parsing |
| Data Validation | Pydantic | Type safety |
intelligentOne provides five core atomic tools that combine into infinite possibilities:
| Tool | Purpose | Example |
|---|---|---|
listen_to_rss() |
Monitor RSS feeds | Track TechCrunch AI news |
extract_ogp_capabilities() |
Extract metadata | Get article details |
search_internal_db() |
Query knowledge base | Find existing workflows |
calculate_competitive_delta() |
Analyze gaps | Compare capabilities |
send_alert() |
Notify users | Urgent updates |
- ✅ MCP-native architecture
- ✅ Atomic tools implementation
- ✅ Hypothesis Engine with LLM generation
- ✅ Simulation Lab with LLM-as-Judge
- ✅ Blueprint Vault persistence
- ✅ Dynamic composite tool loading
- 🔄 Multi-model ensemble judgment
- 🔄 A/B testing of workflow variants
- 🔄 Automatic hyperparameter tuning
- 🔄 Cross-blueprint learning
- 🔄 Failure analysis and self-healing
- 📋 Multi-user Blueprint Vaults
- 📋 Distributed execution engine
- 📋 Advanced security sandboxing
- 📋 Real-time collaboration
- 📋 Enterprise integrations (Slack, Teams, Email)
- Architecture Deep Dive - Technical implementation details
- Getting Started Guide - Step-by-step tutorial
- API Reference - Tool and function documentation (coming soon)
- All workflow hypotheses execute in simulation sandbox before deployment
- Blueprint Vault uses JSON file storage (no SQL injection risk)
- MCP protocol provides built-in authentication
- LLM API keys stored in environment variables (never committed)
- Optional approval threshold configuration
We welcome contributions! intelligentOne is about expanding the universe of autonomous intelligence capabilities.
MIT License - See LICENSE file for details.
intelligentOne represents the future of AI systems:
Before: Humans design workflows, AI executes them
After: AI designs, tests, and deploys workflows autonomously
This is not incremental improvement—it's a fundamental shift in how intelligent systems evolve. While traditional systems degrade over time requiring manual updates, intelligentOne gets smarter with every query, building an ever-expanding library of proven capabilities.
The platform that owns this self-evolution loop will dominate the next decade of AI automation.
Welcome to autonomous intelligence.