Business Context Shapes Agentic Analytics

Organizations are moving toward agentic analytics, in which AI agents investigate business questions, weigh options, recommend decisions and can carry tasks through to action. Their answers depend on shared definitions, governed data and open standards; companies need to teach agents how their business is structured, what its terms mean and how its software connects.

Business intelligence has moved through four eras: IT-built reports, self-service tools, augmented analytics using machine learning, and generative AI copilots and agents. Gartner’s 2016 Magic Quadrant said business-user-centric tools had reached “a tipping point”; in 2021, Gartner said platforms were differentiating on augmented analytics, including machine-assisted data preparation, query generation and insight generation. Copilots accept plain-language questions, while agents can follow up, investigate anomalies and trigger responses.

Gartner predicts that by 2028, at least 15% of day-to-day work decisions will be made autonomously through agentic AI, up from zero percent in 2024. It also forecasts that 33% of enterprise software applications will include agentic AI by 2028, compared with less than one percent in 2024. Agents can take work from a question to an action, while specialists review steps involving modeling, data preparation or visualization.

Companies can supply business context through an ontology (a model of business entities and their relationships) and a semantic layer (definitions of metrics, synonyms and instructions). Gartner predicts semantic layers will be critical infrastructure alongside data platforms and cybersecurity by 2030. In September 2025, Snowflake, Salesforce, dbt Labs, RelationalAI and BlackRock launched the Open Semantic Interchange, a vendor-neutral specification for data definitions.

The article anticipates analytics being built into employees’ existing tools as 2027 approaches. Its example has a Finance lead ask a question in a chat window, an agent check governed definitions and run the analysis, and a workflow open for approval, with humans monitoring the process. Data scientists would continue building custom models and tailoring reports while agents handle routine investigations.

Leaders are advised to plan for slow returns and project cancellations. MIT’s NANDA initiative reported in The GenAI Divide - State of AI in Business 2025 that roughly five percent of AI pilots among the organizations studied reached rapid revenue acceleration, citing a learning gap and flawed enterprise integration. Gartner expects over 40% of agentic AI projects to be canceled by the end of 2027 because of rising costs, unclear business value or inadequate risk controls. It also flags “agent washing” and estimates only about 130 of thousands of agentic AI vendors are genuine.

Gartner projects that by 2030, 50% of AI agent deployment failures will stem from inadequate runtime enforcement on AI governance platforms. Suggested controls include permissions, audit trails, usage and cost tracking, and sandbox testing before production. The article also cites Deloitte’s 2025 agreement to repay the final installment of a contract with Australia’s Department of Employment and Workplace Relations after its report was found to contain fabricated citations, warning that automatically triggered agents could put similar errors into live transactions.

Companies are urged to decide who owns business context: data engineers can manage deployment, credentials, integrations and performance, while business experts define what metrics such as “customer,” “revenue” and “active” mean. dbt Labs’ Joey Gault says the people who best understand these concepts are typically not data engineers and that there is no universal answer to who should own the semantic layer. For example, Sales may count revenue when a contract is signed, while Finance recognizes it after obligations are met. Give each definition a distinct name, document its owner, filters and grain, decide which one answers an unqualified revenue question, and assign a named tiebreaker, typically Finance.

The article recommends versioning every metric, naming its owner, keeping a change history and notifying downstream agents and dashboards before definitions change. Logging which agents ran, who used them, what data they accessed and what teams spent can help trace incorrect answers to their definitions. Gartner expects half of organizations to use AI—

Organizations are moving toward agentic analytics, in which AI agents investigate business questions, weigh options, recommend decisions and can carry tasks through to action. Their answers depend on shared definitions, governed data and open standards; companies need to teach agents how their business is structured, what its terms mean and how its software connects.

Business intelligence has moved through four eras: IT-built reports, self-service tools, augmented analytics using machine learning, and generative AI copilots and agents. Gartner’s 2016 Magic Quadrant said business-user-centric tools had reached “a tipping point”; in 2021, Gartner said platforms were differentiating on augmented analytics, including machine-assisted data preparation, query generation and insight generation. Copilots accept plain-language questions, while agents can follow up, investigate anomalies and trigger responses.

Gartner predicts that by 2028, at least 15% of day-to-day work decisions will be made autonomously through agentic AI, up from zero percent in 2024. It also forecasts that 33% of enterprise software applications will include agentic AI by 2028, compared with less than one percent in 2024. Agents can take work from a question to an action, while specialists review steps involving modeling, data preparation or visualization.

Companies can supply business context through an ontology (a model of business entities and their relationships) and a semantic layer (definitions of metrics, synonyms and instructions). Gartner predicts semantic layers will be critical infrastructure alongside data platforms and cybersecurity by 2030. In September 2025, Snowflake, Salesforce, dbt Labs, RelationalAI and BlackRock launched the Open Semantic Interchange, a vendor-neutral specification for data definitions.

The article anticipates analytics being built into employees’ existing tools as 2027 approaches. Its example has a Finance lead ask a question in a chat window, an agent check governed definitions and run the analysis, and a workflow open for approval, with humans monitoring the process. Data scientists would continue building custom models and tailoring reports while agents handle routine investigations.

Leaders are advised to plan for slow returns and project cancellations. MIT’s NANDA initiative reported in The GenAI Divide - State of AI in Business 2025 that roughly five percent of AI pilots among the organizations studied reached rapid revenue acceleration, citing a learning gap and flawed enterprise integration. Gartner expects over 40% of agentic AI projects to be canceled by the end of 2027 because of rising costs, unclear business value or inadequate risk controls. It also flags “agent washing” and estimates only about 130 of thousands of agentic AI vendors are genuine.

Gartner projects that by 2030, 50% of AI agent deployment failures will stem from inadequate runtime enforcement on AI governance platforms. Suggested controls include permissions, audit trails, usage and cost tracking, and sandbox testing before production. The article also cites Deloitte’s 2025 agreement to repay the final installment of a contract with Australia’s Department of Employment and Workplace Relations after its report was found to contain fabricated citations, warning that automatically triggered agents could put similar errors into live transactions.

Companies are urged to decide who owns business context: data engineers can manage deployment, credentials, integrations and performance, while business experts define what metrics such as “customer,” “revenue” and “active” mean. dbt Labs’ Joey Gault says the people who best understand these concepts are typically not data engineers and that there is no universal answer to who should own the semantic layer. For example, Sales may count revenue when a contract is signed, while Finance recognizes it after obligations are met. Give each definition a distinct name, document its owner, filters and grain, decide which one answers an unqualified revenue question, and assign a named tiebreaker, typically Finance.

The article recommends versioning every metric, naming its owner, keeping a change history and notifying downstream agents and dashboards before definitions change. Logging which agents ran, who used them, what data they accessed and what teams spent can help trace incorrect answers to their definitions. Gartner expects half of organizations to use AI—