Define the concept
Create clear, shared definitions for core business concepts such as customer, product, account and agreement.
Approaches / Data, Meaning and Decisions
Making the organization’s shared language visible.
Data gains meaning through its business concepts, relationships, rules and sources. A semantic layer connects information across systems to a shared enterprise context, enabling people, analytics and AI to work with the same language.
In 2025, Gartner predicted that 60% of AI projects unsupported by AI-ready data would be abandoned by the end of 2026; 63% of organizations were also unsure whether their data management practices were fit for AI.[1]
A customer, account, product or agreement may appear under different names across systems. Enterprise context does more than place them in a shared glossary; it makes their meaning, roles and responsibilities visible.
How is meaning built?
Create clear, shared definitions for core business concepts such as customer, product, account and agreement.
Show how concepts relate, which roles they assume and the business context in which they gain meaning.
Connect meaning to source systems, data ownership, policies, rules and conditions of use.
Use and evolve the shared language across reporting, processes, decisions and AI applications.
A banking example
The same person may be an account holder for one account, a borrower in a credit agreement and a guarantor in a guarantee relationship. The identity remains; the role changes with the business context.
The identity remains the same.
A role is determined not by the person alone, but by the business relationship in which that person participates. The same principle applies to products, agreements, payments, policies and other enterprise concepts.
Shared industry language
Widely accepted industry definitions for concepts such as customer, account, product, agreement, collateral and payment provide a strong starting point. True enterprise context emerges when this shared language is connected to the organization’s processes, ownership, systems and policies.
More than a technical layer
A semantic layer does more than organize data; it makes visible what the organization knows, where that knowledge comes from and under which conditions it can be used.
Technology blueprint of the semantic layer
The semantic layer operates between agents and enterprise information sources: it translates intent into context, reaches the right source and returns results together with meaning, permissions and provenance.
Plans, routes and distributes work across agents.
Agents may access the semantic layer directly; when needed, an orchestrator uses the same layer on behalf of multiple agents.
It is not a single physical data store, but a shared context and access layer operating across multiple sources.
Forrester defines the next-generation semantic layer not as a BI add-on, but as a control plane that centralizes metric definitions, calculations and access policies and delivers them consistently to analytics applications and AI agents.[3]
Core applications, CRM, ERP and operational systems
Data warehouse · data lake · lakehouse
Policies, procedures, agreements and other enterprise content
IoT, telemetry, event streams and external data services
Seeing the whole does not mean building everything at once. Drawing the full picture, then starting with one critical concept and one decision, provides a more robust path in which each proven step informs the next; the customer example above is a natural starting point.
Business and technology teams use the same concepts with the same meaning, making reports and decisions comparable.
Definitions, relationships, rules and provenance are preserved independently of individuals, so knowledge is not lost through change.
Analytics and AI can access not only data, but its meaning and context of use, making outcomes more explainable.
Gartner also frames the context layer required by agents through three components: semantics, operational state and provenance.[2]
Enterprise outcome
The outcome is not merely better-organized data, but an organization that adapts to change, can explain its decisions and supplies AI with trusted context.
If you would like to discuss the details, you can send a note. · LinkedIn ↗
Sources