Underneath the charts is a data plane, and it is a product in its own right, not a detail of the analytics screen. It connects the sources, decides whether a question is answered at the source or from the platform's own storage, holds the description of what the company's numbers mean, and executes the query next to the data.
Everything that shows a figure reads through it: a dashboard, a page of an app, a spreadsheet, an AI agent. That is the mechanical reason those four cannot show four different numbers.
Where to go next
- Sources and storage - what can be connected, when data is queried in place and when it is kept in the built-in data lake, and how freshness is controlled.
- The data model - the semantic layer, how a dataset is enriched into a model, and the join traps that silently inflate a number in every other reporting tool.
- AI in the data layer - what the assistant does with data and why its output is a versioned repository of requirements instead of a generated query.
- Under the hood - the architecture end to end: what executes a query, why Rust, DuckDB and Arrow, caching and backpressure, and where access control is applied.
What reads from all this is Analytics.