.svg)
PRODUCT UPDATE • ADOC 26.7.0
The critical question you're already asking about agentic AI is whether the data means what the agent thinks it means. "Revenue" in Finance and "Revenue" in Sales are not the same column. "Customer" in your CRM and "Customer" in your warehouse may not represent the same entity. Without semantic context, your agents aren't consuming trusted data — they're consuming labeled rows.
You could tolerate that when a human analyst sat between the data and the decision — they could sense-check a definition, ask a colleague, reconcile the inconsistency. An autonomous agent acting at scale cannot. Business Glossary is how you give agents and analysts a shared, authoritative interpretation of your data estate — not defined once at setup, but kept current as your estate and your business evolve.
How Business Glossary Works
Business Glossary organizes your terminology into a three-level hierarchy — Glossaries, Categories, and Terms — where each Term carries a definition, related terms, synonyms, and direct associations to the catalog assets and Data Products that implement it.
Semantic completeness isn't a one-time classification exercise. It's only useful if the meaning your agents and analysts rely on reflects how the business actually uses the data today — not how it used it when someone last ran a governance initiative.
For Data Products specifically, this means the semantic layer and the packaging layer are connected at both ends. A Data Product carries the business terms that describe what it contains. An agent querying that product through the SDK gets something that is not just typed and governed — it is interpreted. The glossary is what makes the product legible to a consumer who doesn't know the underlying schema.
What's New in 26.7.0
Term Relationships: Linked, Not Duplicated
Every data estate accumulates synonyms, near-synonyms, and domain-specific variations of the same concept. "Customer" and "Client" mean different things in different contexts — but without a formal relationship between them, both exist as isolated definitions that compound confusion rather than resolve it. Business Glossary makes term relationships first-class: you can define parent/child hierarchies (Revenue → Gross Revenue, Net Revenue), related terms (Customer, Account, Client), and explicit synonyms that map directly to the canonical term without creating a duplicate definition.
For an agent navigating your estate, linked relationships are navigational paths. A query that resolves "Client" to "Customer" via a synonym mapping surfaces the right assets without requiring the agent to know that your organization uses both terms depending on the source system. The glossary becomes a semantic resolution layer — not just a repository of definitions your team has to already know exist.
Steward Workflows: Lifecycle Management at Scale
Creating terms is only part of the governance work. Keeping them current — retiring outdated terms, merging duplicates, reassigning ownership as teams change — is where governance programs typically break down. Business Glossary gives stewards a structured workflow for each of these operations. Terms move through a defined lifecycle. Ownership is role-based. Every operation is audit-logged. Merging a duplicate term reconciles all its associations in the background without requiring manual remediation across individual assets.
Most governance programs solve for classification at a point in time. Terms get defined during a data initiative, then drift out of sync as the estate grows and teams change. A steward workflow with lifecycle management and background reconciliation means your semantic layer stays current without a separate reconciliation project every quarter.
What This Looks Like in Practice
Onboarding a New Data Domain
A governance team onboarding a new Payments domain creates a Glossary, organizes terms into Categories — Transaction, Settlement, Compliance — and defines each term with its domain-specific meaning. As the data engineering team crawls Payments tables, stewards associate newly discovered assets with the relevant terms. By the time the first Data Product is published from those assets, it already carries the business context consumers need to evaluate and trust it. The semantic layer arrives with the product, not after a separate documentation exercise.
Resolving a Synonym Conflict Across Source Systems
A data steward discovers that "Account" in the CRM maps to "Client" in the data warehouse and "Customer" in the analytics layer — three terms, one concept, no formal relationship. Rather than maintaining three separate definitions, the steward establishes "Customer" as the canonical term, links "Account" and "Client" as synonyms, and associates all three source assets with the canonical definition. Downstream consumers — and agents — now resolve all three to a single authoritative meaning. The reconciliation happens once; every downstream association inherits it.
An Agent Navigating to Trusted Data by Business Term
An agentic AI system tasked with retrieving data relevant to "Customer Retention Rate" resolves the term through the Business Glossary, navigates to the associated Data Product, and confirms the product's quality status before consuming it. The agent didn't need to know the underlying table name, the schema it lives in, or which database it belongs to. The glossary is the entry point that makes the path to trusted data deterministic — not dependent on how well the agent knows your physical estate.
Who Benefits
- Data stewards — a structured workflow to create, maintain, retire, and merge terms without manual cross-system remediation; every operation audit-logged
- Governance leads and CDOs — a canonical term library that replaces ad-hoc labelling and gives compliance teams a defensible, lifecycle-managed record of business definitions
- Data product owners — business context attached to their products at publish time, so consumers understand what the product contains without a separate documentation exercise
- Data consumers and analysts — navigate to the right dataset by business term, without needing to know table names or schema structures
- Agentic AI systems — a semantic resolution layer that makes data products queryable by business meaning, not just by technical identifier
Getting Started
- Navigate to Governance → Glossary and create your first Glossary, scoped to a domain your team already owns.
- Add Categories to reflect the key concept groups within that domain.
- Define Terms — start with the five to ten terms that cause the most confusion or inconsistency across teams today.
- Associate each Term with the catalog assets and Data Products that implement it.
- Assign steward ownership and establish a review cadence for term lifecycle management.
For full configuration guidance, term relationship schema, and association management, see the Business Glossary documentation at docs.acceldata.io.
Acceldata • Data Observability Platform • acceldata.io
