Text2Dashboard: A Governed Agent Architecture for Enterprise Natural-Language Dashboards

It is a DataBrain-specific prototype that turns natural-language analytic requests into inspectable dashboards, using an installable Codex plugin and a standalone Agent Runtime that combines schema-constrained model decisions with typed tools, persistent state, and deterministic Hooks for approval, audit, checkpointing, recovery, and failure handling.

The architecture separates the model, which proposes actions, from deterministic software that controls execution and records state transitions, enforcing read-only SQL plus static checks, dynamic preflight, and browser inspection, rather than relying directly on raw model output.

No. The paper reports small, DataBrain-specific results and the authors explicitly state they do not establish production readiness, general text-to-SQL accuracy, or an efficiency advantage over manual dashboard construction.

They measure the behavior of approval, audit, checkpointing, recovery, and failure-handling Hooks; all ten scenarios met their specified outcomes without unapproved external side effects.