Summit - Autonomous Intelligence System

Summit turns raw data into explainable intelligence in minutes. It unifies intelligence analysis, deterministic graph databases, and LLM-powered context assembly into a single operational loop.

🏆 GA Status: v5.6.0-GA

Summit has successfully passed the General Availability (GA) Certification Contract.

• Integrity: 100% Machine-Verifiable (Zero-Mocks Architecture)
• Security: O(1) Groth16 SnarkJS recursive verification (Admissibility Gate)
• Compliance: Continuous Ledger mapping against SOC 2 and NIST AI RMF

To run the authoritative GA certification audit locally:

pnpm run ga:verify:gate

✨ Features

• 🤖 Agentic AI: Multi-agent orchestration for autonomous research
• 🕸 Knowledge Graphs: Neo4j + GraphRAG for connected intelligence
• 📡 Real-time Ingest: Streaming connectors for CSV, S3, REST APIs
• 🔍 Vector Search: Semantic retrieval with embeddings
• 📈 CompanyOS SDK: Enterprise intelligence APIs
• 🔒 Security Hardened: Production-ready CORS, Helmet, Zero-Knowledge gates

⚡ See It In Action (The 90-Second Wow Demo)

Watch me go from zero to full OSINT report on any company in 47 seconds — fully local, provenance guaranteed.

Run the Demo Locally

pnpm demo:company
# Or directly: ./scripts/wow.sh

This single command:

1. Auto-ingests the MIT Sloan Startups dataset via Switchboard
2. Triggers semantic chunking and GraphRAG multi-hop traversal
3. Deploys the Maestro swarm (Jules + Codex + Observer)
4. Outputs a full provenance report with confidence scores and source links

---

🚀 Quickstart

Prerequisites

• Node.js 20+
• Docker & Docker Compose
• Neo4j 5.x (via Docker)
• PNPM (v10+ enforced)

Install & Run Locally

To physically spin up the secured local production-twin infrastructure:

pnpm install
pnpm build
docker-compose --file docker-compose.hardening.yaml up -d

Server runs at http://localhost:4000

🌟 Golden Path (Recommended)

For a deterministic and clean bring-up of the local development environment:

# Automates: make clean -> make bootstrap -> make up
./scripts/golden-path.sh

Or manually:

make clean      # Clean build artifacts and docker system
make bootstrap  # Install Python venv and Node dependencies
make up         # Start all services via Docker Compose

First Query

# GraphQL playground
curl -X POST http://localhost:4000/api/graphql \
  -H "Content-Type: application/json" \
  -d '{"query": "{ health { status version } }"}'

# Or use the web UI
open http://localhost:5173

⚖️ Governance & Compliance

Summit is engineered for high-stakes, regulated environments where safety, accountability, and explainability are non-negotiable.

• Governance Framework: The end-to-end enforceable architecture for safety-first autonomy and continuous compliance.
• Evidence & Audit: How we prove compliance via cryptographic signatures, the Provenance Ledger, and automated control mapping.
• Summit Constitution: The canonical doctrine defining the absolute constraints on agentic action, human primacy, and the purpose of the platform.

🏛 Architecture

Summit follows a modular microservices architecture:

┌──────────────────────────────────┐
│ 👥 User Agents (Jules, Codex)   │
└──────────┬───────────────────────┘
           │
           v
┌──────────┬────────────────────────┐
│ API Layer │ GraphQL + REST APIs    │
└──────────┬────────────────────────┘
           │
   ┌───────┼────────┐
   │       │        │
   v       v        v
┌────────┐ ┌────────┐ ┌────────┐
│ Ingest │ │GraphRAG│ │ Agents │
│ Engine │ │Pipeline│ │ Orchest│
└────┬───┘ └───┬────┘ └───┬────┘
     │         │          │
     v         v          v
┌─────────────────────────────────┐
│ 📊 Data Layer                   │
│ Neo4j | Postgres | Redis | Qdrant│
└─────────────────────────────────┘

🧩 Core Components

• IntelGraph: Central graph data model for entities, relationships, evidence, and provenance to enable link analysis and multi-hop queries.
• Maestro Conductor: Workflow orchestration for job DAGs, retries, policy enforcement, observability, and SLO tracking, driven by IntelGraph state.
• CompanyOS: Knowledge, runbooks, and business logic APIs integrated with graph data and orchestrated workflows.
• Switchboard: Ingestion, normalization, deduplication, enrichment, and routing of events into the platform's core services.
• Provenance Ledger: Evidence-first audit trail for explainable outputs, lineage, and compliance-ready reporting across the platform.
• Operational Memory System: Persistent operational intelligence that remembers, learns, and reuses execution experience (see Memory as the Moat below).

💾 Operational Memory System (Memory as the Moat)

Summit implements a persistent operational memory infrastructure that transforms the platform from a stateless execution engine into a learning system that compounds intelligence over time.

Core Principle

decision = retrieval + adaptation (not pure generation)

Instead of generating decisions from scratch each execution, the platform:

1. RETRIEVES similar past executions from the Verifiable Transformation Ledger (VTL)
2. ADAPTS execution strategies based on historical experience
3. RECORDS outcomes to expand the memory moat

Strategic Value

Traditional systems generate decisions stateless, with no improvement over time. Memory-backed systems reuse proven strategies and compound intelligence, creating a defensible competitive advantage.

---

📚 Key Components

Connectors

Ingest data from multiple sources:

• REST APIs: Poll external services
• CSV/S3: Batch file processing
• Neo4j/Postgres: Database replication
• Webhooks: Real-time event streaming

GraphRAG

Retrieval-augmented generation with knowledge graphs:

• Entity extraction & linking
• Multi-hop graph traversal
• Vector similarity search
• LLM-powered synthesis

Agents

Autonomous AI agents for research and analysis:

• Jules: PR reviewer, code analyzer
• Codex: Task brief generator
• Observer: Telemetry and monitoring

📡 API Reference

• GraphQL Schema
• REST Endpoints
• CompanyOS SDK

🧪 Testing

# Run all tests
pnpm test

# Run deterministic GA Gate Tests
pnpm run ga:verify:gate

🛡 Security

Summit follows security best practices:

• Zero-Knowledge cryptographic recursive proofs (SnarkJS)
• Helmet.js for HTTP security headers
• CORS with explicit origin whitelisting
• Rapid Depedency Patching (0 Critical/High CVEs)

Demo Flow

Check out the Demo Script to walk through a pilot demonstration.

License

Proprietary