Letting people ask business questions in plain English (natural-language query, or text-to-SQL) is useful. But by itself it is not enough. The real opportunity is to rethink the analytics workflow, moving from pull-based reporting to push-based decision support.
From pull to push
In the old model, users had to find the dashboard, interpret the metric, decide whether the signal mattered, and translate it into action.
In the new model, the system should surface the exception, explain the driver, recommend the next best action, and let the user follow up in plain language.
Two kinds of expert knowledge
That only works if two different kinds of subject-matter knowledge are built into the system.
Finance and corporate knowledge defines what the numbers mean: KPI definitions, metric trees, thresholds, business rules, variance logic, and what counts as a meaningful exception. It helps the system understand what matters, not just what changed.
Operational knowledge defines what to do when something changes: root-cause patterns, field constraints, escalation paths, action playbooks, and what actually works in the real world. It helps the system move from explanation to action.
The bottleneck moved upstream
None of this works without a strong, unglamorous foundation: clean tables, clear metric definitions, readable documentation, good joins, date logic, null handling, business rules, and enough context for the system to know what question it is really answering.
So the human job does not vanish. It shifts upstream. The AI writes the query. People make the data understandable enough for the query to be right.
And if you want agents to act inside business workflows, not just answer questions, a semantic layer is still only part of the context they need. They also need to know which real-world customer or asset is meant, how customers, contracts, assets, locations and people relate, what changed and when, and what the AI is allowed to do next.
Where the value comes from
Plain-English querying is only the interface. The real value comes from analytics agents curated by subject-matter experts, combining trusted metric logic with operational decision logic. Finance and corporate experts help the system understand the business. Operational experts help it act in the business.
That is how analytics moves from passive reporting to active decision support.
Originally posted on LinkedIn on May 19, 2026, lightly edited. See the original post and discussion.