Before an agent uses data, help it understand the data.
A table name is not enough to make a sound decision. Let people and agents discover assets, inspect schemas and find business definitions, with lineage and quality context to guide the next question.
A field guide to the data behind the answer.
Use catalog tools to discover connected assets and inspect their structure. Follow descriptions and lineage to understand what a dataset represents and where it comes from, then use semantic tools to ask a governed data question.
Review configured quality checks and freshness history alongside the asset. Keep measured evidence distinct from unmeasured quality, and use the source permissions required for the task.
From “which table?” to a defensible next step.
Illustrative task: prepare a renewal review. Discover the relevant contract and account assets, inspect the relationship between them, and look for published definitions of renewal measures.
Use configured semantic tools to work with approved definitions. Execute against supported sources with the person’s permissions, then carry the result into a review or report.
Agentic discovery. Deliberate authority.
Agents can use catalog and semantic tools to discover available context and ask governed data questions. People retain the work of reviewing definitions, publishing approved meaning and deciding what a task may use.
Connection method, supported dialect, network access, credentials and permissions are part of the design. The catalog describes data; it is not a new copy of all source rows.
Evaluate one business question, end to end.
Start with a connected source and a real question. Inspect the assets, descriptions and lineage that help a person or agent select the right inputs.
Add a quality check, review a business definition and inspect an authorized query. Then repeat with a person who has different access.
Bring the dataset people keep asking you to explain.
Let’s walk through discovery, meaning, quality and access using a question that matters to your team.