A catalog is only as useful as what is written in it. SelectZero’s AI enrichment drafts that writing for you. Descriptions for catalog objects and their attributes, related glossary terms, and values for custom fields. It runs on the objects you select in the catalog, across connections, and nothing is applied until a person has reviewed each suggestion.

The problem

Most catalogs start out as a list of names. After the first metadata sync there are thousands of tables and columns, and almost none of them say what they hold. The description on fct_payments.amount is empty. Nobody has linked customer.email to the glossary term that defines it. The PII flag your data protection officer asked for exists as a custom field, but it has been filled in for a few dozen columns out of several thousand.

Filling that in by hand is weeks of steward time, and it rarely finishes before the next source is connected. Until it does, searching by business term returns little, and the catalog is not context anyone can rely on, whether that is an analyst or an AI assistant.

What enrichment does

For each selected object and its attributes, enrichment suggests:

• Descriptions for the object and for every attribute.
• Related glossary terms, drawn only from the terms already defined in your business glossary.
• Values for custom fields, for example a domain on a table or a PII or CDE flag on a column, for whichever custom fields your organisation has defined.

Suggestions are built from context the platform already holds: technical metadata, profiling and sampled values, and related glossary terms. Every suggestion is shown next to the current value and is reviewed and applied one at a time. A steward can accept a description, keep the existing domain and drop one of two proposed terms in the same pass.

Use cases

• Onboarding a new source. A data engineer connects a new warehouse schema, selects its tables and runs enrichment. Stewards start from a reviewed draft instead of an empty page.
• Flagging critical data elements. A governance lead wants every column in mart_credit_exposure marked as CDE or not. Enrichment proposes a value per column, the steward confirms or corrects it, and the flag becomes searchable across the catalog.
• Connecting the glossary to physical data. The term Customer is defined but linked to nothing. Running enrichment across the warehouse and CRM connections suggests it on the matching columns, so the term’s Assets tab shows where the concept actually lives.

How to run it

1. In the Catalog view, select the objects to enrich in the catalog and click “Enrich”
   • Or open an individual object click “Enrich”
2. Define the fields to enrich and add context if needed
3. Review each suggestion and compare against the current value.
4. Apply the selected suggestions.

If you want to try it out, start with one schema nobody has documented yet. The review queue will quickly show how much writing your stewards no longer start from scratch, and how much your glossary is carrying. If the suggestions come back thin, that tells you where to invest next.