CHECKLIST, CONDITION 2 OF 10, read 13 Nov 2026
Documented: 11 of 17 tools check it by rule, 5 leave it to a person
- The test
- A description or business definition exists for the asset and its key columns.
- Why it matters
- Agents work from names and descriptions. With only a schema, an agent guesses what a column means and returns a confident wrong answer.
- How to check it
- Read the asset page or the model YAML. Check that the table and the columns the agent will touch each have a description that says what they mean in business terms.
- What failing looks like
- A column named amt_2 has no description, and the agent treats it as net revenue when it is gross.
What each use case needs
| Use case | Need | Why |
|---|---|---|
| Metric agent | required | The agent picks columns by their documented meaning, so missing descriptions become wrong answers. |
| Coding agent | required | A coding agent reads descriptions to learn what a model is for before it changes the logic. |
| Write-back agent | recommended | Write-back targets are usually fixed fields, but a description helps the agent map values to the right field. |
11 tools check documented by rule
- OvalEdge, rank 1, 64.5.
- Euno, rank 2, 62.5.
- DataHub, rank 3, 61.0.
- OpenMetadata, rank 4, 61.0.
- Atlan, rank 5, 59.0.
- Secoda, rank 6, 58.5.
- dbt, rank 8, 52.5.
- Alation, rank 9, 50.5.
- Qlik Talend Cloud, rank 11, 46.5.
- Actian, rank 15, 39.0.
- Informatica, rank 16, 36.0.
5 tools leave documented to a person
- Databricks Unity Catalog, rank 7, 58.0.
- Ataccama, rank 10, 49.5.
- Snowflake Horizon, rank 13, 45.5.
- Precisely, rank 14, 43.0.
- Collibra, rank 17, 36.0.
1 tools show none found on 13 Nov 2026
- Monte Carlo, rank 12, 46.0.
What the sources say
- Atlan, AI-ready data for talk-to-data: "schema alone scored 9% correct or close; adding table descriptions, clusters and example queries lifted it to 49%."
- dbt Labs, 5 dbt MCP server patterns that work in production: "The model reads the lineage graph, drafts model, and column descriptions, and you review them in the PR."
- Gartner, Lack of AI-Ready Data Puts AI Projects at Risk (Q&A with Roxane Edjlali, Feb 26, 2025; archived copy): "Proving AI readiness of the data is a process and a practice based on the availability of metadata to align, qualify and govern the data."
- McKinsey, A data leader's technical guide to scaling gen AI (archived copy; mckinsey.com did not load from this run): "Organizations that use gen AI-augmented data catalogs can facilitate real-time metadata tagging, including automatically generating metadata from structured and unstructured content and creating smart tags."
Other conditions
Sources
- Atlan: AI-ready data for talk-to-data, read 13 Nov 2026
- dbt Labs: 5 dbt MCP server patterns that work in production, read 13 Nov 2026
- Gartner: Lack of AI-Ready Data Puts AI Projects at Risk (Q&A with Roxane Edjlali, Feb 26, 2025; archived copy), read 13 Nov 2026
- McKinsey: A data leader's technical guide to scaling gen AI (archived copy; mckinsey.com did not load from this run), read 13 Nov 2026
Last reviewed 13 Nov 2026.