Why this matters
Missing lineage, stale profiles and undocumented assets each point at a different agent. Knowing which agent produces which metadata is how you find the cause, and how you decide what runs, how often, and at what cost.
Before you start
Different agents do different jobs
Ingestion, lineage, profiling, classification and documentation are separate concerns handled by separate agents. When something is missing from Collate, identifying which agent owns that output is the fastest route to the cause.
Scheduling is a cost and freshness decision
Every agent run consumes resources on the source system as well as in the platform. Frequency should follow how fast the underlying thing changes. Schema changes slowly, data volume changes constantly, and they do not need the same cadence.
Failed agent runs are silent
A failed agent run produces no visible error for consumers, only metadata that is a little more out of date each day. Checking agent runs is a recurring task.
The interactive demo
Work through the demo below. It runs in your browser with nothing to install. Watch for:
- An overview of each of the metadata agents
- An overview of the AI agents
This demo is also available on its own at the Learning Center.
After this module you should be able to
- Name which agent is responsible for which part of the platform
- Diagnose a metadata gap by identifying the agent that should have filled it
- Choose a schedule that balances freshness against cost