What is business context for AI?

Summary

• Business context for AI is the set of authoritative definitions, metrics, relationships, and domain knowledge that lets AI understand and answer questions accurately on enterprise data.
• Without it, AI fills gaps with inference, using the wrong tables, misinterpreting concepts, and producing generic or incorrect answers, because business context is scattered across dashboards, queries, pipelines, wikis, tickets, documents, and chat.
• Databricks provides business context through Genie Ontology, a unified context layer that combines Unity Catalog semantics you define and govern with context Genie infers automatically from your assets and usage.
• Unity Catalog semantics include metric views (governed KPI definitions), domains, a business glossary, and certification signals, so agents retrieve trusted, relevant context instead of searching the entire catalog.
• Genie ranks the most authoritative definition, resolves conflicts, and answers only from sources you are permitted to see, so everyone works from a single, trusted set of definitions.

What is business context for AI?

Business context for AI is the set of authoritative definitions, metrics, relationships, and domain knowledge that enables AI to understand and answer questions accurately on enterprise data. Without it, an AI system lacks the ground truth it needs to interpret data correctly and instead fills the gaps with inference, producing generic or incorrect answers. The core challenge is that this context is scattered across dashboards, queries, pipelines, wikis, tickets, documents, and chat threads, so when AI cannot easily find it, it guesses. On Databricks, business context is captured and served through Genie Ontology, built on Unity Catalog.

Why AI needs business context

To answer enterprise questions reliably, an AI agent needs several kinds of context:

• Shared definitions. How the organization defines key terms and metrics, such as "revenue," "active customer," or "qualified lead."
• Asset relevance. Which data assets matter to different areas of the business.
• Trusted sources. Which sources can be relied on for accurate answers.
• Domain knowledge. The business rules, relationships, and hierarchies that connect concepts.

Without this context, agents may use the wrong tables, misinterpret important business concepts, or overlook the assets that matter most. For example, asked to forecast revenue, an agent without context might search the entire data estate instead of recognizing the Finance domain and drawing on the curated, high-quality assets there.

Why Databricks Genie Ontology

Genie Ontology is a unified context layer that brings together two categories of context so Genie can ground its answers.

• Unity Catalog semantics (modeled context you define and govern). Unity Catalog semantics capture context you author and control: metric views are reusable SQL objects that define and govern business KPIs; domains and subdomains group assets into business-aligned categories so agents retrieve relevant context instead of searching the entire catalog; glossary pages provide authoritative definitions of business concepts and taxonomies; and certification and deprecation signals mark which assets are trusted or outdated.
• Inferred context (extracted and maintained automatically). Genie also infers context from your environment, extracting snippets from dashboards, SQL queries, notebooks, pipelines, and Genie Agents, including metric definitions, authoritative sources, and business rules. Each snippet receives an authority score based on where it was generated, how often it is used, and how fresh it is.
• Authoritative ranking with OntoRank. When you ask a question, Genie ranks the most relevant snippets using OntoRank, a PageRank-style authority score that selects the most authoritative definition when definitions conflict, resolves those conflicts, and answers using only sources you are permitted to see. Grounding in this context improves accuracy and reduces latency by narrowing the data Genie needs to crawl and query.
• Metric views as the highest-authority signal. Metric views capture how a measure is calculated and are defined once, so a term like "GMV" means the same thing on a Genie Agent, on a dashboard, or in a third-party agent, giving the organization a single source of truth for its metrics.

Together, the human-modeled Unity Catalog semantics and the automatically inferred context form the unified context layer that Genie draws on to answer questions in the organization's own language. See Introducing Genie One, Genie Ontology, and Genie Agents.

Getting started

• Read Introducing Genie One, Genie Ontology, and Genie Agents for the concept and architecture.
• Learn how modeled context works in Unity Catalog semantics and define governed KPIs with metric views.
• See how domains build enterprise context for AI and review open and adaptive enterprise context.

FAQs

What is business context for AI?

It is the authoritative definitions, metrics, relationships, and domain knowledge that let AI interpret enterprise data correctly. Without it, AI infers meaning and produces generic or incorrect answers.

Why do AI agents give wrong answers without business context?

Because the context they need, definitions, trusted sources, and domain rules, is scattered across dashboards, queries, wikis, tickets, and documents. When an agent cannot find it, it guesses, which leads to hallucinations and misused tables.

What is Genie Ontology?

Genie Ontology is a unified context layer on Databricks that combines Unity Catalog semantics you define and govern with context Genie infers automatically from your assets and usage, so Genie can ground answers in trusted business context.

How do metric views provide business context?

Metric views are reusable SQL objects that define and govern how a KPI is calculated. Because a metric is defined once, it means the same thing across Genie, dashboards, and third-party agents, creating a single source of truth.

How does Genie choose the right definition when definitions conflict?

Genie uses OntoRank, a PageRank-style authority score, to rank context by where it came from, how often it is used, and how fresh it is, then selects the most authoritative definition and answers only from sources you are permitted to see.