Data manager · DataHub developer
Shipped (V0)Business rule generation from plain language
Write 'flag NAV deviations above 2%' — the generator returns a syntactically valid business rule as a typed object DataHub consumes directly.
Implemented in the platform catalog and running in task mode. The agent's DataHub tools are production-grade contracts, contract-tested and fixture-backed until live engine connectors land in Beta.
In DataHub
The daily reality
The work this replaces
Business-rule authoring is slow, expert-bound and backlogged. The syntax lives in a handful of heads and a thousand pages of documentation, so every new control waits on the same few specialists.
How it works
What the agent actually does
dh-br-generator
Spec in plain language
A data manager describes the rule — 'flag NAV deviations greater than 2%' — and DataHub submits it to the dh-br-generator agent as a typed task.
Grounded drafting
The agent drafts the expression using the BR syntax dictionary and a corpus of real examples, so the rule respects the engine's actual grammar — not a generic model's guess at it.
Typed result, not prose
The draft returns as a schema-validated object the product consumes directly. If the output violates the schema, the run fails hard — never silent success.
Chains into the quality pipeline
From there, the DataHub workflow can chain validation, evaluation against assertions and test generation — each step a governed agent run.
The outcome
Outcome
Rules drafted in minutes against the real syntax dictionary and rule corpus, instead of waiting in the specialists' backlog.
Governed by design
The generator is declared in the Git-governed catalog, pinned to its model, role-gated (DH_DEVELOPER) and carries a versioned eval baseline that can only go up.
FAQ
Questions teams ask about this
What input does the generator need?
A plain-language specification of the control you want, submitted by DataHub as typed input. The agent grounds its drafting in the BR syntax dictionary and an examples corpus, and returns the rule as a schema-validated object.
Can the agent change rules in production?
No. It drafts and returns a typed object; what happens next belongs to the DataHub workflow and its owners. Task agents are stateless and bounded, with no side effects beyond their declared, policy-controlled tools.
Is there a conversational version?
Yes — dh-br-assistant, an interactive BR expert embedded in the rule-authoring screen, is in Beta (V1). It drafts, explains, validates and tests rules through conversation; any action touching the live engine shows a confirmation card first.
Go further
Related use cases and concepts
Business rule validation & testing
Shipped (V0)On every rule save, DataHub chains governed agents: validate syntax and best practice, evaluate against business assertions, build and run tests.
Data-quality remediation
In product · unified at GAData-quality errors investigated in the context of the feeds, rules and configuration involved — so remediation becomes review-and-apply.
Data onboarding & configuration Q&A
Beta (V1)'How do I configure a Kafka feed?' — answered from documentation and live configuration objects, with deep links to the exact screens.
From the glossary
Join the Early Adopter Program
General availability lands in Q3 2026. The Early Adopter Program is open now — a limited cohort, a one-year platform trial, and three workshop streams: Business ROI, Compliance, and Operational fit.