BI & Analytics Glossary

Agentic analytics is an approach to business intelligence where an AI system pursues a goal or open-ended business question by planning multiple analytical steps, executing them against governed data, and adjusting the plan based on what each step returns. It is built for higher-level, goal-oriented prompts like "why did pipeline drop last quarter," where the next move depends on what the previous step revealed, not just precise lookups like "revenue yesterday in Berlin."

Natural-language analytics lets users ask data questions in ordinary language and receive charts, tables, explanations, or follow-up options without writing SQL. In trustworthy BI, natural-language analytics is grounded in a semantic layer so business terms map to approved metrics and governed data.

Text to SQL is the technique of having a large language model translate a natural-language question directly into a SQL query, with no governed semantic layer in between. It is straightforward to prototype, but in enterprise BI it tends to produce inconsistent metrics, fragile queries, and hallucinated tables or columns, because the model has to infer business meaning from raw schema.

Conversational analytics is a BI interaction model where users explore data through a back-and-forth dialogue instead of a fixed report or one-off query. It lets users ask follow-up questions, refine filters, drill into drivers, and preserve context across the analytical conversation.

Self-service BI is an analytics model where business users can answer their own data questions, build reports, and explore metrics without waiting for a specialist to create every query. AI-native self-service BI uses natural language and governed semantic definitions to make that access easier while keeping data teams in control.

An AI BI assistant is an artificial intelligence interface that helps users ask questions, explore metrics, create visualizations, and automate analytical work inside a business intelligence platform. A trustworthy AI BI assistant is grounded in governed semantic definitions rather than loose guesses about raw data.

Traditional BI usually centers on dashboards, reports, and manual exploration paths, while AI BI adds natural-language interaction, assisted analysis, automated workflows, and conversational follow-up. The best AI BI keeps the governance and reliability expected from traditional BI while making analytics easier for more users.

A semantic layer is a governed business vocabulary that maps raw data structures to concepts, metrics, relationships, and permissions that people and applications can reuse. For AI BI, the semantic layer gives an assistant a controlled representation of the business so it does not have to infer meaning from table names alone.

A knowledge graph for BI is a semantic model that represents business concepts, metrics, relationships, synonyms, and governance rules as connected knowledge. It helps analytics tools and AI assistants understand how a company talks about its data instead of relying only on physical database structure.

Semantic modelling is the work of defining business concepts, metrics, relationships, synonyms, hierarchies, permissions, and calculation rules on top of raw data. It gives BI tools and AI assistants a shared map of what the business means by its data.

An AI BI hallucination is a confident, fluent analytics answer that is not actually grounded in the underlying data or in the company's approved business definitions. The term comes from large language models, and in BI it most commonly arises when an LLM is given the freedom to generate SQL directly or to narrate a result without verified inputs.

Embedded analytics brings dashboards, charts, metrics, or analytical workflows directly into another application, portal, or customer-facing product. It helps users make data-driven decisions in the place where they already work instead of sending them to a separate BI tool.

AI data alerts are proactive notifications that monitor business data and surface important changes, thresholds, anomalies, or recurring insights without requiring users to check dashboards manually. In governed BI, alerts should use approved metrics, permissions, and context so recipients understand why they were notified.

What-if analysis in BI lets users change assumptions or scenario inputs and see how business metrics may respond. It connects reporting with planning by helping teams compare possible outcomes using governed data and reusable business logic.

Forecasting in BI uses historical and current business data to estimate future values for metrics such as revenue, demand, churn, inventory, or operational capacity. In AI BI, forecasts are most useful when they are connected to governed definitions, explanations, alerts, and what-if analysis.