An evolution of the Open Semantic Interchange standard moves beyond BI reporting to the layer of meaning that enterprise AI agents actually need.
Six months ago, we joined Snowflake and a coalition of industry leaders to launch the Open Semantic Interchange (OSI) - an open, vendor-neutral standard for how business semantics get expressed, shared, and trusted across the data and AI stack. The promise was simple: define a metric, or a dimension once, and have every tool, dashboard, and AI agent in your stack understand it the same way.
That first draft of the spec codified the dimensional models that BI tools live on. But as enterprises started building agentic systems on top of their data clouds, one observation came back again and again: dimensional models support analytics and reporting, but agents need to reason.
Today, the OSI community is taking the next step. We are extending OSI spec to support ontological semantics alongside its dimensional foundation - because it ensures broad interoperability and preserves existing enterprise investment in the data stack, while giving agents the precise, conflict-free vocabulary they need to act with confidence
A dashboard can tell you that revenue by region = SUM(amount) GROUP BY region. That's enough to render a chart on a dashboard. It is not enough for an agent to act.
Ask an agent: "If our Tier-2 supplier in Shenzhen goes offline next month, which finished goods and customer commitments are exposed?" It immediately hits ambiguity. What counts as "exposed" - any presence in the bill of materials, or only above a content threshold? Which vendors qualify as Tier-2 - sub-suppliers of our direct suppliers, or any vendor two hops upstream? And should the trace stop at finished goods, or follow the chain forward to open orders, contractual commitments, and the customer accounts behind them?
Dimensional models can roll up shipments by region or aggregate spend by supplier. They can't tell you what a Tier-2 supplier actually is, which entities qualify, or how to traverse the chain: supplier → parts → sub-assemblies → finished goods → orders → customers to assemble an answer that spans multiple hops across relationships.
Such questions need agents to understand meaning and identity of organizational concepts, relationships between them, and any constraints - these are the core elements of an ontology. Without it, every agent rediscovers business logic from scratch on every turn - making it expensive, slow, and unreliable.
Anchoring OSI in ontological structures delivers four concrete benefits for the enterprise:
Cross-ecosystem interoperability. Enterprises that have already invested in semantic technologies - across BI, data catalogs, knowledge graphs, and application platforms - don't have to choose between ecosystems or rebuild work to participate in OSI. The same definitions move across tools.
Trusted AI grounding. An ontology informs agents about the business: its entities, relationships, and rules. This grounds decisions in business logic, making them more trustworthy and accurate. It also results in a far lower token bill, as agents avoid reasoning from scratch on every query.
Durability against change. Business meaning shouldn't get rewritten every time a pipeline is refactored or a schema evolves. Anchoring semantics at the conceptual level gives the enterprise a stable, reusable contract that survives infrastructure churn.
Executable business logic. Rules and constraints become computable, not just documented, enforceable in data validation, queryable as part of the model, and consumable by reasoning systems.
The net effect: define your semantics once, and use them everywhere — from a dashboard to a decision agent — with confidence that everyone in your stack means the same thing.
This work is the product of the OSI Ontology Representation Working Group: a cross-industry team spanning RelationalAI and data platforms, BI vendors, governance providers, and large financial enterprises that came together to pressure-test the proposal across BI, ontology, and applied-AI lenses. The spec is stronger because of that collective effort, and we're grateful to every contributor.
Start building with the Rel API, install the native app inside Snowflake, or get in touch about access to our ready-to-use Rel experience.
Rel is a decision agent aligned to your business, grounded in your semantic model, and powered by the advanced reasoners of the RelationalAI decision intelligence platform.
©2026 RelationalAI. All rights reserved.
