ATLANTA, June 02, 2026 (GLOBE NEWSWIRE) -- Today, Airia announced major new capabilities for its Context Engineering solution, which powers building accurate, governed AI applications on enterprise data. The release includes Graph RAG with customizable knowledge graphs, a Semantic Layer for domain vocabulary grounding, and a Knowledge MCP Server for multi-hop agentic retrieval.
Context Engineering has always been the foundation of effective AI. These capabilities make it work with the messy, complex, permission-controlled data that real enterprises actually have.
Answering questions that require connecting the dots
Enterprise AI applications regularly fail on questions that span multiple documents, regulatory frameworks, or product domains. Answering them requires understanding how concepts relate, not just which documents contain similar text.
Airia Context Engineering now includes Graph RAG, which extracts domain-specific entities from documents during ingestion and maps the relationships between them. Teams can define their own entity types and ontologies based on their industry and use case, and refine them over time without re-ingesting data. The result is an AI application that can answer complex, multi-step questions by tracing connections across a document library, finding answers that keyword or vector search alone would miss.
Multi-hop retrieval for complex queries
The Knowledge MCP Server exposes Airia's retrieval capabilities as callable tools for any LLM. When a question requires multiple searches across different document types or indexes to build a complete answer, the model can chain those retrieval steps automatically. This closes a critical gap in enterprise AI: most retrieval systems are built for single-shot lookups. Real enterprise questions are not.
AI that understands how your organization talks
Generic AI models struggle with internal terminology, product codes, and domain-specific language. The Semantic Layer lets teams upload their own vocabulary, and Airia uses it to ground every query, without retraining the model.
Permission-aware retrieval throughout
Every retrieval step enforces access controls. Users only surface content they are authorized to see, with no manual review and no risk of sensitive data appearing in the wrong context.
Airia Context Engineering works with any LLM and deploys as SaaS, private cloud, or on-premises.
"The enterprises getting the most out of AI are the ones investing in what happens before the model sees the data," said Kevin Kiley, CEO of Airia. "Airia Context Engineering gives teams the tools to make that layer accurate, domain-aware, and trustworthy."
About Airia
Founded in 2024, Airia delivers the industry’s first unified enterprise AI security, orchestration, and governance platform, purpose-built to accelerate AI adoption. Airia guides the world’s most innovative enterprises through their AI transformation journey by addressing the critical gap between rapid innovation and governance requirements—empowering teams to build and deploy AI agents fast while maintaining enterprise-grade control. Learn more at airia.com.
Media Contact:
Julia Harold
juliaharold@airia.com
