Agents / Data Governance Office
Data Steward Agent
Watches every lakehouse data product on each load for freshness, quality rules and schema change in Dataplex, keeps lineage current, recomputes the Data Health Score, and tickets the data owner before a downstream model is affected. Publishes the new version when nothing flags; holds it when the PII scan does.
An agent executes the process end to end and logs what it did. No human is in the path, because there is no discretion to exercise and no consequence to carry.
- Acts for
- Data Steward
- Action class
- write
- Decisions / 30d
- 96
- Safe-state returns / 30d
- 4
Processes covered
- Dataset load and profiling
- Data quality, lineage and schema watch
- Steward ticketing
Stages:Monitor & revalidate
Memory
Retains schema history per dataset for change detection; no row-level retention.
Safe state
Holds the incoming load unpublished and tickets the steward when profiling or the PII scan cannot complete.
Reviewer time per case: 0 min with this agent, 115 without.
Ownership and sensitivity
Who approves access
- Owner, Data Governance Office, Ahmad Faizal
Business-sensitive. Models and data products scoped to named business units.
Entitlement per business unit, approved by the owner; conditions attach.
Tools
- Load watcher
- Dataplex data quality and profiling
- Dataplex lineage
- PII and customer-identifier scanner
- Data Health Score calculator
- Ticketing
Data access scope
- Read-only: incoming loads and existing schemas
- Write: health scores, new dataset versions, steward tickets
- PII scan results are kept as counts and column names, never as values
Guardrails
- A PII flag holds the version for the data owner's sign-off; it is never auto-published
- Schema changes that break a consuming model open a ticket to both the steward and the model owner
- Never deletes or overwrites a prior version
Human-in-the-loop
- The data owner signs off when the PII scan flags anything
Orchestration
Every tool call is scoped by the declared data access above; the orchestrator cannot reach systems outside it. Owning business unit: Group Data & AI.
Live demo run
Watch the agent execute a real scenario step by step, every tool call, validation, and human checkpoint is traced and auditable. Typical run: ~5.2s.
Used in workflows
Retiring this asset would require these chains to be re-pointed first.
Audit
Updated 2026-07-31. Run traces retained 24 months for audit under the platform governance policy.
Community · 0 threads
Questions, findings and requests from the business units that use this asset. Owners reply here; threads with upvotes surface to the owning team's inbox.