OpenMetadata Monthly Meetup - Community Spotlight: "Your Data, Your Agents, One Context Platform"

Slides from OpenMetadata's September Community Meetup, held Wednesday, September 23, 2026. The deck covers two sessions. Paul Phillips of GingerHawk Solutions, Sr Data Architect, shows how OpenMetadata becomes a trust layer for data-naive clients, walking through a demo built around an arcade business with a data agent that answered questions about check-ins, revenue, and attendance, followed evidence lineage to trace every answer back to its source, restricted the agent to gold-certified datasets, and closed with agent use classification analytics and a Ledger MCP that caches evidence so the agent does not spend tokens recreating the same lookups. Harsha of Collate, Ceo & Co-founder, gives a first look at Ontology Studio, the centerpiece of the upcoming OpenMetadata 2.1 release, distinguishing a glossary from a taxonomy, an ontology, and a knowledge graph, then defining concepts with properties and relationships, mapping them to data assets, and querying them through SPARQL or a visual builder, with import and export in OWL, RDF, CSV, and JSON-LD and one-click startup packs for FIBO, FHIR, and GS1. Ontology Studio ships in the 2.1 release on November 3rd.

Catch the next OpenMetadata Community Meetup @ https://luma.com/collate

Transcript

1. Community Meeting Sep 2026
2. Star us on GitHub https://github.com/open-metadata/OpenMetadata Join our Slack https://slack.open-metadata.org/ Subscribe on Youtube https://www.youtube.com/OpenMetadataChannel Discover Collate SaaS https://www.getcollate.io/
3. Agenda • Welcome to our Community Spotlight • Paul walks us through his use case for OMD • Infrastructure topology view, demoing what OMD enables • Harsha gives use more of 2.1 for Collate and OMD • Metrics and Community Shoutouts • Q&A 3
14. OPENMETADATA 2.1 · CONTEXT PLATFORM Ontology Studio Modeling business meaning so AI agents can reason over your data Glossar y → Ontolog y → Knowledge Graph → Agent s
15. OpenMetadata Context Platform Three primitives for people and agents working with data 01 02 · THIS TALK 03 Context Ontology Memory What data exists: metadata, lineage, quality, policies, access What data means: business concepts and typed relationships What has been settled: approved decisions and corrections
16. What Is an Ontology A formal, machine-readable model of a business domain Layer What it gives you Example Glossary Agreed names and definitions “Churn Rate: share of customers who cancelˮ Taxonomy A single hierarchy Retention → Churn → Voluntary Churn Ontology Typed relationships, properties and rules Churn Rate calculatedFrom Cancelled Subscriptions Knowledge graph The ontology plus real tables, owners and lineage subscriptions.cancelled_at → Churn Rate → exec_kpis 👉 The ontology is the blueprint; the knowledge graph is the building
17. From Definition to Model Churn Rate, as a glossary entry and as an ontology concept GLOSSARY TERM ONTOLOGY CONCEPT “Churn Rate: the share of customers who cancel their subscription in a given period.ˮ isA Readable by people. Opaque to software. Metric calculatedFrom Active Subscriptions, Cancelled Subscriptions partOf Retention attributes unit = percentage · grain = monthly exactMatch fibo-fnd:CustomerAttritionRate Computable: impact, derivation and equivalence are data.
18. WHY ONTOLOGIES MATTER NOW LLMs are extremely good at language and extremely bad at knowing which table you meant.
19. Disambiguation, Not Retrieval “What was revenue last quarter?ˮ SIMILARITY SEARCH FINDS 14 THE ONTOLOGY KNOWS tables with a revenue column All plausible. No way to tell which one is authoritative. → Recognized Revenue is the metric Finance signs off on It is calculated from Booked minus Deferred Revenue revenue_v2_final is deprecated
20. Graph-Shaped Questions Enterprise questions are multi-hop, not similarity lookups “Which dashboards break if we drop this column?ˮ column → lineage → dashboard “Everything classified PII that flows outside the EUˮ tag → lineage → region “Tier-1 tables with no owner and no data contractˮ tier → owner? → contract? 👉 The answer is a set of edges, not a generation
21. Advantages for Organizations What a governed model of meaning gives the business One definition per concept Guardrails agents can read Finance, Sales and Marketing resolve “revenueˮ to the same governed term Axioms and relationship rules become a contract that can be validated Semantics outlive pipelines Computable impact analysis Tables get renamed and warehouses migrate; the concept Customer does not Changing a definition shows every metric, asset and team it touches
22. How Ontologies Enable Agentic AI Agents traverse the graph instead of guessing 01 · Ask 02 · Resolve 03 · Traverse 04 · Answer Question Ontology Knowledge graph Explainable result A user or agent asks in business language The term maps to one governed concept and its rules Follow edges to certified tables, owners and lineage Every step is a path through the graph that can be shown 👉 Exposed to agents through knowledge-graph MCP tools on the Context Platform
23. OPENMETADATA 2.1 · CONTEXT PLATFORM Whatʼs Shipping in 2.1 Ontology Studio and the knowledge graph
24. One Workspace, Four Modes Author, explore and query the business ontology in one place View Navigate concepts and trace them to data Edit Model concepts, relationships, attributes, axioms Query Ask graph questions in a SPARQL console AI An assistant that reasons over the graph
25. Built on the Glossary You Already Have Every ontology construct is a versioned, governed entity Ontology concept In OpenMetadata Standard Ontology Glossary owl:Ontology Concept Glossary Term owl:Class Typed relationship Relationship Type owl:ObjectProperty Attribute Ontology Attribute owl:DatatypeProperty Logical axiom Ontology Axiom owl:disjointWith, … External crosswalk Concept Mapping skos:exactMatch, …
26. Standards-Aligned Starter Packs Install an industry ontology, then adapt it to your business FIBO HL7 FHIR GS1 Financial services Healthcare Retail and product data EPCIS ISA95 HR Open Supply chain events Manufacturing Human resources 👉 Or import existing OWL files to reuse enterprise taxonomies
27. The Knowledge Graph Projection The whole catalog, mirrored into RDF as a derived index SYSTEM OF RECORD PROJECTION UNLOCKS Relational store RDF triple store Catalog-wide SPARQL Concepts, assets, owners, lineage stay authoritative → Apache Jena Fuseki, indexed like search → Reasoning and SHACL validation Graph insights and 3D view Knowledge-graph MCP tools 👉 Beta and off by default; authoring and exploration work without it
28. Getting Started From glossary to agent-ready knowledge graph 01 Model Author concepts in Ontology Studio or install a starter pack 02 Attach Assign terms to tables, columns, dashboards, pipelines 03 Project Enable RDF and index the catalog into the knowledge graph 04 Query Use the SPARQL console or connect agents via MCP
29. Thank You Search tells an agent what exists. Lineage tells it what flows. An ontology tells it what things mean.
30. Community Metrics and Updates
31. OpenMetadata Community Growth 494 15,303 14,235 53,800 31
32. Weekly Community Metrics 1010 PRs in the last 4 weeks 232 Qs in the last 4 weeks 32
33. Community Contributions stout-yeoman fmcardoso zerafachris Shreyansh100704 Vansh0310 zhang-arvin XXUCHAN jfmlima larspettermadsstuen DaniS2001 33
34. Q&A