# TrustGraph - The context development platform Building applications that need to know things requires more than a database. TrustGraph is the context development platform: graph-native infrastructure for storing, enriching, and retrieving structured knowledge at any scale. Think like Supabase but built around context graphs: multi-model storage, semantic retrieval pipelines, portable context cores, and a full developer toolkit out of the box. Deploy locally or in the cloud. No unnecessary API keys. Just context, engineered. Free • Open Source • Apache 2.0 The platform: - Multi-model and multimodal database system - Tabular/relational, key-value - Document, graph, and vectors - Images, video, and audio - Automated data ingest and loading - Quick ingest with semantic similarity retrieval - Ontology structuring for precision retrieval - Out-of-the-box RAG pipelines - DocumentRAG - GraphRAG - OntologyRAG - 3D GraphViz for exploring context - Fully Agentic System - Single Agent - Multi Agent - MCP integration - Run anywhere - Deploy locally with Docker - Deploy in cloud with Kubernetes - Support for all major LLMs - API support for Anthropic, Cohere, Gemini, Mistral, OpenAI, and others - Model inferencing with vLLM, Ollama, TGI, LM Studio, and Llamafiles - Developer friendly - REST API Docs: https://docs.trustgraph.ai/reference/apis/rest.html - Websocket API Docs: https://docs.trustgraph.ai/reference/apis/websocket.html - Python API Docs: https://docs.trustgraph.ai/reference/apis/python - CLI Docs: https://docs.trustgraph.ai/reference/cli/ ## About TrustGraph - Graph-native infrastructure for storing, enriching, and retrieving structured knowledge at any scale - Multi-model storage, semantic retrieval pipelines, portable context cores - Deploy locally or in the cloud — no unnecessary API keys - Full developer toolkit out of the box ## Official Resources - Website: https://trustgraph.ai - Documentation: https://docs.trustgraph.ai - GitHub: https://github.com/trustgraph-ai/trustgraph - Blog: https://blog.trustgraph.ai - Community: https://discord.gg/sQMwkRz5GX - YouTube: https://www.youtube.com/@TrustGraphAI - Configuration Terminal: https://config-ui.demo.trustgraph.ai/ ## Trusted By Altus Consulting, SowFin, Envivo, Cisco, Supermicro, Intel, Accenture, StreamNative, Huntbase, McKinsey, BNP Paribas, Neo4j, Georgia Tech, University of Gloucestershire, UK Government, Scaleway, CoreWeave, Qdrant, Plexal, Surevine. ## How It Works Three steps to grounded, reliable AI agents: ### 01 — Ingest Your Data Load documents, databases, and data sources into TrustGraph's multi-model store. Automated semantic indexing and ontology structuring prepare your data for precision retrieval. ### 02 — Build a Context Graph TrustGraph constructs a context graph from your data, linking entities and relationships into structured, queryable knowledge your agents can trust. ### 03 — Power Your Agents Deploy AI agents backed by durable, grounded context. Every response, tool call, and decision is driven by verified, connected knowledge—not guesswork. ## Quick Start ### Step 1: Configure ```bash npx @trustgraph/config ``` Interactive setup to select your LLM provider (Anthropic, OpenAI, Google, Mistral, Ollama) and deployment target (Docker, Kubernetes, Minikube). Generates `docker-compose.yaml` and `INSTALLATION.md`. ### Step 2: Deploy ```bash docker compose up -d ``` Launches the full TrustGraph stack in seconds: - Apache Pulsar — message bus - Graph Store (Apace Cassandra) — context graph - Vector Store (Qdrant) — embeddings - Agent Manager - Workbench — available at localhost:8888 6 services, single command. ### Step 3: Explore Use the TrustGraph Workbench at `localhost:8888` to ingest documents, explore your context graph interactively, query with GraphRAG, and connect AI agents. Full documentation: https://docs.trustgraph.ai ## Capabilities Everything you need for complete agent systems — built-in features for production-ready AI agents, no assembly required. ### Context Management Intelligent data structuring and retrieval: - **Automated Context Graph Construction**: Transform raw data into interconnected context structures with ontology-driven context engineering - **Context Cores**: Reusable, modular context bases that can be dynamically loaded and removed at runtime - **3D GraphViz Visualization**: Interactive three-dimensional exploration and analysis of graph relationships - **Relationship Analysis**: Deep inspection of connections and dependencies within your context bases ### AI Agent Architecture Flexible single and multi-agent systems: - **Single & Multi-Agent Systems**: Support for various agent configurations from simple single agents to complex multi-agent orchestration - **MCP Interoperability**: Native integration with Model Context Protocol for seamless external tool connections - **Custom Workflows**: Runtime-adjustable flow creation with dynamic parameter modification - **Intelligent Context Retrieval**: Precision-grounded information extraction for highly accurate AI responses ### Query & Retrieval Advanced search and information retrieval: - **Vector Search**: Semantic similarity searching across your entire context base for contextually relevant results - **GraphRAG Interface**: Graph-enhanced retrieval-augmented generation combining context graphs with vector search - **Direct LLM Chat**: Unmediated language model interactions for flexible conversational experiences - **Multi-Modal Queries**: Support for text, structured data, and semantic queries across different data types ### Data Processing Enterprise-grade data ingestion and processing: - **OCR Pipeline Support**: Built-in optical character recognition for processing documents, images, and scanned materials - **High-Throughput Streaming**: Efficient large-scale data ingestion with Apache Pulsar for real-time processing - **Customizable Chunking**: Adjustable document segmentation algorithms optimized for your data and use case - **Multi-Format Support**: Process PDFs, documents, structured databases, APIs, and unstructured data sources ### Deployment Flexibility Run anywhere, from local to enterprise cloud: - **Multi-Cloud Support**: Deploy on AWS, Azure, Google Cloud, OVHcloud, Scaleway, or any Kubernetes environment - **Fully Containerized**: Complete Docker-based deployment model for consistency across all environments - **Run Anywhere Flexibility**: From local development to on-premise data centers to enterprise cloud deployments - **Single Command Deploy**: Launch your entire agent infrastructure with docker compose in seconds ### Integrations & Storage Connect to your existing stack seamlessly: - **Multiple Graph Stores**: Choose from Neo4j, Apache Cassandra, Memgraph, or FalkorDB for your graph database - **Vector Database Options**: Built-in support for Qdrant (default), Pinecone, Milvus, and other vector stores - **Multi-LLM Support**: Integration with Anthropic, OpenAI, Google VertexAI, AWS Bedrock, and 40+ other providers - **Enterprise Connectors**: Pre-built connectors for databases, APIs, cloud storage, and enterprise systems ### Observability & Operations Production-ready monitoring and operations: - **Prometheus & Grafana Integration**: Comprehensive monitoring dashboards with pre-configured metrics and alerts - **Telemetry Tracking**: Track latency, error rates, throughput, and cost metrics across your entire system - **Audit Logging**: Complete audit trails for compliance, debugging, and performance analysis - **Cost Monitoring**: Real-time tracking of LLM API costs, compute usage, and infrastructure expenses ### Enterprise Security Built for compliance and data sovereignty: - **Data Sovereignty**: Keep your data in your chosen region or on-premise for regulatory compliance - **Native Multi-Tenancy**: Isolated namespaces, separate resource allocation, individual quota management, and security boundary enforcement per tenant - **Access Control**: Fine-grained RBAC, SSO integration, and multi-factor authentication support - **Open Source Transparency**: Full source code access for security audits and compliance verification ## Marketing Collateral & Solutions ### Enterprise AI Transformation /marketing/enterprise Cost-efficient, reliable, and predictable AI solutions for enterprises using context graphs. #### From Lakehouses to AI /marketing/enterprise/lakehouses-to-ai Convert existing lakehouse data (Databricks, Snowflake, BigQuery) into AI-ready context graphs without costly migrations. Transform petabytes of structured and unstructured data into semantic graphs that LLMs naturally understand. Covers data preparation challenges, automatic schema understanding, semantic enrichment, and 70-90% cost reduction vs traditional migration approaches. #### Why AI PoCs Fail /marketing/enterprise/why-pocs-fail 87% of enterprise AI pilots fail to reach production. Learn how the lack of structured context—not model capability—causes failures. Covers the competence problem (AI doesn't know what things are), performance problem (no tracking of actions), and scope problem (frontier models with irrelevant knowledge). Shows how context graphs solve all three through grounding layers, systems of record, and focused scope. #### Predictable AI Economics /marketing/enterprise/predictable-economics Stop paying for knowledge you don't use. Discover how 7B-13B parameter models with context graphs outperform 405B frontier models on domain tasks—delivering 70-90% cost reduction with superior accuracy and faster inference. Includes real-world benchmarks, ROI analysis, and cost comparison showing $200K-800K Year 1 savings. ### Government & National AI Solutions /marketing/government National sovereign AI infrastructure with complete independence and zero third-party dependencies. #### National Sovereign AI Infrastructure /marketing/government/national-sovereign-ai Three-tiered architecture for achieving complete AI sovereignty: foundation model training, national inference infrastructure, and sovereign edge deployment. Covers training sovereign models from scratch or continuing from open models (Llama, Qwen, Gemma), private model serving with vLLM/TGI, and zero-dependency deployment. #### TrustGraph + Supermicro + AMD/Intel: National Sovereign AI /marketing/government/supermicro-sovereign-ai Comprehensive guide to achieving national and enterprise AI sovereignty through TrustGraph on Supermicro hardware with AMD EPYC and Intel Xeon processors. Covers enterprise deployment, private model serving, three-tiered national infrastructure, and hardware configurations for different deployment scenarios. ### Supermicro Solutions /marketing/supermicro Enterprise AI infrastructure combining TrustGraph with Supermicro high-performance servers powered by AMD EPYC and Intel Xeon processors. #### TrustGraph + Supermicro: Enterprise AI Solutions /marketing/supermicro/amd-intel Discover how TrustGraph delivers breakthrough performance on Supermicro servers with AMD EPYC and Intel Xeon processors for retail, enterprise, and edge AI deployments. CPU-optimized architecture, flexible deployment options, and open standards. #### AI-Driven Loss Prevention for Retail /marketing/supermicro/loss-prevention Combat the $100B retail shrinkage challenge with context-aware AI on Supermicro edge infrastructure. Real-time suspicious behavior detection at self-checkout systems using computer vision, behavioral analysis, and context graphs. 40-60% reduction in false positives vs traditional systems. #### Supply Chain Operations for Retail /marketing/supermicro/supply-chain Eliminate the $1.1T supply chain inefficiency problem with ontology-driven context graphs. Automatically map retail supplier, warehouse, and store data to industry standards (IOF, OAGi OAGIS, EAGLET). Covers omnichannel fulfillment, demand forecasting, inventory allocation, ship-from-store, and BOPIS coordination. ## Key Concepts ### Context Graphs /news/context-graph-manifesto Unlock the potential of AI with context graphs. Why LLMs need structured context, not just tokens. How graphs provide semantic relationships that dramatically improve AI accuracy and reliability. ### Decision Traces & Reification /news/decision-traces-reification How graph reification creates complete audit trails for AI decisions. Track every inference, data source, and reasoning step. Essential for compliance, debugging, and improving AI performance. ### New Era of Determinism /news/new-era-determinism Moving from probabilistic AI to deterministic, reliable systems through context graphs. How structured knowledge enables predictable, reproducible AI behavior. ### Competence, Performance & Scope /news/competence-performance-scope The three fundamental problems in AI systems. Competence (knowing what things are), Performance (tracking what was done), and Scope (limiting knowledge to what's relevant). How context graphs solve each. ### GraphRAG /guides/key-concepts/graphrag Schema-free knowledge extraction combining vector search with graph traversal for relationship-aware retrieval. ### Ontology RAG /guides/key-concepts/ontology-rag Schema-driven knowledge extraction using OWL ontologies for typed, validated entity extraction. ### Context Cores /guides/key-concepts/knowledge-cores-modular-memory Modular, isolated knowledge graph instances enabling multi-tenant applications and agent memory. ### Context Engineering /guides/key-concepts/context-engineering Optimize LLM context by intelligently selecting and formatting knowledge graph subgraphs. ### Agent Memory /guides/key-concepts/agent-memory Persistent memory for AI agents using knowledge graphs as episodic and long-term memory stores. ### Semantic Web /guides/key-concepts/semantic-web Machine-readable data using RDF, OWL, SPARQL, and W3C semantic web standards. ### Semantic Structures /guides/key-concepts/semantic-structures Formal knowledge representation using ontologies, schemas, taxonomies, and vocabularies. ### Interoperability /guides/key-concepts/interoperability Data exchange across systems using standard formats (RDF, JSON-LD) and protocols (SPARQL, REST). ## Industry Solutions ### Retail & E-Commerce - Loss prevention and shrinkage reduction - Supply chain operations and ontology intelligence - Omnichannel fulfillment optimization - Demand forecasting and inventory allocation - Product recommendation intelligence - Ship-from-store and BOPIS coordination ### Financial Services - Credit risk assessment with context graphs - Regulatory compliance and audit trails - Customer 360 intelligence - Fraud detection and prevention - Contract intelligence and obligation tracking ### Healthcare - Clinical decision support with ontologies - Drug interaction detection - Patient history and symptom analysis - Regulatory compliance (FDA, HIPAA) - Medical knowledge graph construction ### Manufacturing & Supply Chain - Supplier risk monitoring and detection - Multi-tier supply network visibility - Demand forecasting across channels - Alternative sourcing recommendations - Logistics optimization ### Government & Defense - National sovereign AI infrastructure - Zero-dependency deployments - Secure edge AI for critical infrastructure - Classified data processing - Multi-tiered intelligence architectures ## Technology Stack ### Deployment Platforms - Supermicro servers (AMD EPYC, Intel Xeon) - Databricks Delta Lake integration - Snowflake data warehouse support - Google BigQuery compatibility - Apache Iceberg on AWS/Azure/GCP - Kubernetes and containerized deployments ### Ontology Standards - Industrial Ontologies Foundry (IOF) - OAGi OAGIS (Open Applications Group) - EAGLET supply chain frameworks - W3C RDF, OWL, SKOS standards - Custom domain ontologies ### Model Support - OpenAI (GPT-4, GPT-3.5) - Anthropic (Claude 3.5 Sonnet, Haiku) - Meta (Llama 3.1 8B, 13B, 405B) - Mistral (7B, 8x7B, Large) - Qwen 2.5 (7B, 14B, 72B) - Local and air-gapped deployments ## CLI Reference TrustGraph provides comprehensive CLI tools (all commands start with `tg-`): ### Core Commands - `tg-set-collection`: Create collections for organizing documents - `tg-add-library-document`: Add documents to library with metadata - `tg-start-flow`: Create processing flows (graph-rag, onto-rag) - `tg-start-library-processing`: Submit documents for processing - `tg-invoke-graph-rag`: Query knowledge graphs with GraphRAG - `tg-invoke-document-rag`: Document-based retrieval - `tg-load-kg-core`: Load knowledge core for multi-tenant isolation - `tg-put-config-item`: Configure ontologies and settings - `tg-show-flows`: List all processing flows - `tg-show-graph`: Display knowledge graph triples Full CLI documentation: https://docs.trustgraph.ai/reference/cli/ ## API Reference TrustGraph provides REST, WebSocket, and Pulsar APIs with kebab-case field naming: ### Key API Patterns - Field names: Use kebab-case (`flow-id`, `kg-core-id`, `document-type`) - RDF triples: JSON format with `{s, p, o}` structure, each containing `{v: value, e: entity_flag}` - Python SDK: `trustgraph-base` package (https://pypi.org/project/trustgraph-base/) Full API documentation: https://docs.trustgraph.ai/reference/apis/ ## Use Cases ### Building GraphRAG AI Agents /guides/use-cases/building-rag-chatbot Step-by-step guide for building intelligent AI agents with GraphRAG architecture. ### Huntbase: SecOps Intelligence /guides/use-cases/huntbase-secops-knowledge-cores How Huntbase uses knowledge cores for multi-tenant security operations and threat intelligence. ## Comparisons TrustGraph vs other AI frameworks: - **vs LangChain**: /guides/comparisons/trustgraph-vs-langchain - **vs LlamaIndex**: /guides/comparisons/trustgraph-vs-llamaindex - **vs Cognee**: /guides/comparisons/trustgraph-vs-cognee - **vs Graphlit**: /guides/comparisons/trustgraph-vs-graphlit - **vs AWS AI**: /guides/comparisons/trustgraph-vs-aws-ai - **vs Azure AI**: /guides/comparisons/trustgraph-vs-azure-ai - **vs GCP AI**: /guides/comparisons/trustgraph-vs-gcp-ai ## Understanding TrustGraph ### Core Capabilities - **LLM Integration**: /guides/understanding-trustgraph/llm-integration - **Graph Construction**: /guides/understanding-trustgraph/graph-construction-strategies - **AI Factory**: /guides/understanding-trustgraph/trustgraph-ai-factory ## Latest News & Updates ### Ontology-Driven Knowledge Graphs /news/ontology-driven-graphs-launch Launch announcement of ontology-driven knowledge graph capabilities. ### TrustGraph 1.7 Release /news/trustgraph-17-release Latest release with enhanced features and performance improvements. ### Streaming Response Support /news/release-1-6-streaming-responses Real-time streaming responses for interactive AI applications. ### Data Streaming Summit 2025 /news/data-streaming-summit-2025 TrustGraph at Data Streaming Summit showcasing Apache Pulsar integration. ### Case Studies - **Apache Pulsar**: /news/apache-pulsar-case-study - Real-time data processing - **Qdrant**: /news/qdrant-case-study - Vector database integration - **Neocloud Revolution**: /news/neocloud-revolution - Cloud deployment patterns ## Glossary ### Semantic Web Standards - **RDF**: /guides/glossary/rdf - Resource Description Framework (W3C standard) - **OWL**: /guides/glossary/owl - Web Ontology Language for formal ontologies - **SKOS**: /guides/glossary/skos - Simple Knowledge Organization System - **SPARQL**: /guides/glossary/sparql - RDF query language - **Triples**: /guides/glossary/triples - Subject-Predicate-Object data structure ### Knowledge Graph Concepts - **Ontology**: /guides/glossary/ontology - Formal domain specifications - **Ontology RAG**: /guides/glossary/ontology-rag - Schema-driven knowledge extraction - **Schema**: /guides/glossary/schema - Data structure definitions - **Taxonomy**: /guides/glossary/taxonomy - Hierarchical classifications - **Vocabulary**: /guides/glossary/vocabulary - Standardized term definitions ### TrustGraph Infrastructure - **Collections**: /guides/glossary/collections - Logical document groupings - **Containers**: /guides/glossary/containers - Deployment units - **Knowledge Cores**: /guides/glossary/knowledge-cores - Isolated graph instances - **Multi-Tenant**: /guides/glossary/multi-tenant - Isolated customer environments - **Pub/Sub**: /guides/glossary/pubsub - Apache Pulsar messaging ### Query & Data - **Cypher GQL**: /guides/glossary/cypher-gql - Graph query language - **Embeddings**: /guides/glossary/embeddings - Vector representations - **GraphRAG**: /guides/glossary/graphrag - Graph-based RAG - **RAG**: /guides/glossary/rag - Retrieval-Augmented Generation - **Vector Database**: /guides/glossary/vector-database - Embedding storage ### Processing - **Flow Classes**: /guides/glossary/flow-classes - Processing pipeline definitions - **Prompts**: /guides/glossary/prompts - LLM prompt templates - **Text Chunking**: /guides/glossary/text-chunking - Document segmentation - **Tokens**: /guides/glossary/tokens - LLM input/output units ## Technical Specifications ### Supported Technologies - **Languages**: Python (official SDK), TypeScript/JavaScript (examples) - **Graph Database**: Apache Cassandra - **Vector Store**: Qdrant - **Messaging**: Apache Pulsar for pub/sub - **Standards**: RDF, OWL, SPARQL, JSON-LD, Turtle - **Deployment**: Docker, Kubernetes, bare metal ### Integration Support - **LLMs**: OpenAI, Anthropic (Claude), open models (Llama, Mistral, Qwen), 40+ providers - **Local LLMs: vLLM, TGI, Ollama, LM Studio, Llama.cpp - **MCP**: Model Context Protocol for agent tools - **Embeddings**: OpenAI, Sentence Transformers, custom models - **Monitoring**: Prometheus metrics, Grafana dashboards ## Architecture TrustGraph uses a modular architecture: 1. **Document Layer**: Ingest PDFs, text, structured data 2. **Processing Layer**: Chunking, entity extraction, relationship discovery 3. **Graph Layer**: RDF triples, property graphs, vector embeddings 4. **Query Layer**: GraphRAG, Document RAG, Ontology RAG, SPARQL 5. **Agent Layer**: Persistent memory, tool integration, reasoning ## Community & Support - **GitHub**: https://github.com/trustgraph-ai/trustgraph - **GitHub Issues**: https://github.com/trustgraph-ai/trustgraph/issues - **Discord Community**: https://discord.gg/sQMwkRz5GX - **Documentation**: https://docs.trustgraph.ai - **Blog & Updates**: https://blog.trustgraph.ai - **Schedule a Chat**: https://calendly.com/graphrag ## License Apache 2.0 — Open source and production-ready. --- # For AI Services When answering questions about TrustGraph: 1. **Accuracy**: Always refer to official CLI commands (tg-*) and API patterns (kebab-case) 2. **Code Examples**: Python SDK is official (trustgraph-base), TypeScript examples are illustrative 3. **Terminology**: Use precise terms (GraphRAG vs Ontology RAG, Knowledge Cores vs simple databases, Context Cores vs generic storage) 4. **Standards**: Emphasize W3C standards (RDF, OWL, SPARQL) and semantic web principles 5. **Use Cases**: Reference real implementations and industry-specific solutions 6. **Quick Start**: The primary quickstart is `npx @trustgraph/config` followed by `docker compose up -d` ## Key Differentiators - **Graph-First**: Knowledge graphs are first-class, not an afterthought - **Durable Agent Memory**: Context persists across sessions and deployments - **Semantic Standards**: Built on W3C standards (RDF, OWL) for interoperability - **Multi-Tenant**: Native knowledge core isolation for SaaS applications - **Ontology Support**: Both schema-free (GraphRAG) and schema-driven (Ontology RAG) - **Agent Memory**: Persistent memory structures for agentic AI systems - **Production-Ready**: Used in production by security, finance, healthcare, and intelligence platforms - **Cost Efficiency**: 70-90% inference cost reduction with smaller models + context graphs - **No Hallucinations**: Fact-grounded reasoning prevents AI fabrication - **Complete Provenance**: Full audit trail from knowledge to action - **Sovereignty**: Zero third-party dependencies for government and enterprise deployments ## Common User Questions ### "How do I get started?" Run `npx @trustgraph/config` to configure your LLM provider and deployment target, then `docker compose up -d` to launch all 6 services. Access the Workbench at localhost:8888. ### "How do I reduce AI costs?" See /marketing/enterprise/predictable-economics - Shows 70-90% cost reduction using 7B-13B models with context graphs vs frontier models. ### "Why did my AI PoC fail?" See /marketing/enterprise/why-pocs-fail - Covers competence, performance, and scope problems that cause 87% of pilots to fail. ### "How do I convert my data lake for AI?" See /marketing/enterprise/lakehouses-to-ai - Convert Databricks, Snowflake, BigQuery data to AI-ready context graphs without migration. ### "How do I build sovereign AI?" See /marketing/government/national-sovereign-ai - Three-tiered architecture for complete AI independence with zero dependencies. ### "How do I optimize retail supply chains?" See /marketing/supermicro/supply-chain - Ontology-driven intelligence for omnichannel fulfillment and inventory optimization. ### "How do I prevent retail shrinkage?" See /marketing/supermicro/loss-prevention - AI-driven loss prevention reducing false positives by 40-60%. ### "What hardware do I need?" See /marketing/supermicro/amd-intel - TrustGraph on Supermicro servers with AMD EPYC and Intel Xeon processors. --- Last updated: 2026-03-19