TrustGraph Launches Automated Ontology-Driven Graph Building for Reliable AI Agents

TrustGraph Launches Automated Ontology-Driven Graph Building for Reliable AI Agents

TrustGraph - Eliminate AI Hallucinations

SAN FRANCISCO, CA – November 24, 2025 – TrustGraph announces the release of significant new capabilities and infrastructure improvements in release 1.5.8. The latest version introduces support for ontology-driven graph construction, dynamic embeddings model selection, MCP authentication support, and full Python 3.13 compatibility—alongside critical stability enhancements and workbench improvements.

Introducing Ontology-Driven Graph Construction
The centerpiece of this release is a new ontology-based extraction processor that fundamentally changes how organizations extract knowledge from unstructured data. Rather than relying on generic machine learning heuristics, TrustGraph can now leverage domain-specific ontology definitions to guide extraction logic, ensuring that the constructed knowledge graph aligns with business requirements and domain constraints.

"Most knowledge graph platforms ask what can we extract? TrustGraph now asks what should we extract?" said Mark Adams, Co-Founder of TrustGraph. "The ability to automatically build graphs with a specified ontology puts your data fully in control, dramatically reducing noise and ensuring consistency across distributed data exchange. This is what enterprise-grade knowledge graph infrastructure looks like - except TrustGraph builds it autonomously instead of needing a team of data engineers and 6 months."

Key benefits of Ontologies include:

  • Precise Data Extraction: Define extraction rules once in your ontology configuration; apply them consistently across all extraction jobs

  • Entity Context Preservation: Extract entity contexts alongside triple data for richer semantic understanding

  • Workbench Integration: Seamlessly integrates with TrustGraph Workbench's ontology editor for visual rule definition

  • Reduced Post-Processing: Dramatically cuts manual data cleaning and validation downstream

Enterprise-Ready Platform Updates
Beyond ontologies, TrustGraph's latest release brings several enterprise-critical improvements:

Infrastructure & Compatibility:

  • Python 3.13 Support: Full compatibility with Python 3.13, including migration from cassandra-driver to scylla-driver for modern language support

  • Vector Store Lifecycle Improvements: Fixed hard-coded vector size limitations; collections now initialize with proper support for variable embedding dimensions

  • Dynamic Embeddings Model Selection: Users can now select embedding models dynamically, enabling optimization for specific use cases rather than one-size-fits-all approaches

Security & Integration:

  • MCP Authentication: Added MCP (Model Context Protocol) authentication token header support, enabling secure integration with enterprise services and AI platforms

Stability & Reliability:

  • Fixed batch deletion errors that occurred during large-scale collection operations

  • Resolved agent step schema validation issues

  • Improved OpenAI service compatibility for broader LLM integration

  • Comprehensive test coverage for ontology extraction, dynamic embeddings, MCP authentication, and agent step parsing

Enhanced TrustGraph Workbench
The TrustGraph Workbench—the visual interface for designing, testing, and deploying knowledge graphs—also receives substantial improvements:

  • LLM Model Selection UI: Users can now select from multiple LLM providers and models without leaving the workbench

  • MCP Authentication Interface: Enterprise-grade authentication configuration now accessible through an intuitive UI

  • Performance Optimization: System UI performance significantly improved across all workflows

  • Multi-Ontology Support: Fixed critical issues with importing and managing multiple ontologies

  • Crash Stability: Resolved flow class viewer crash that could interrupt design workflows

  • Refactored Architecture: Library breakout and module reorganization improve maintainability and extensibility

Technical Excellence & Documentation
This release includes comprehensive technical specifications for ontologies and vector store lifecycle management, ensuring practitioners have the architectural context needed for production deployments. Full test coverage and updated documentation support enterprise adoption.

Availability
The latest version of TrustGraph is available immediately on GitHub. Comprehensive release notes, technical documentation, and migration guides are available here.

TrustGraph Community Resources