GUIDE, STACK, read 13 Nov 2026
dbt can check five of the ten AI-readiness conditions from its own project files
Five conditions are checkable in dbt, two partly, three not at all
We counted the ten conditions against the dbt Developer Hub pages listed below, read on 13 Nov 2026. Five can be checked from dbt's own files and commands: documented, quality, fresh, lineage and metric definition. Two are partial: owner and access. Three show none found on 13 Nov 2026: certified, sensitivity classified and active use.
The count matters because dbt is where most agent-facing tables are built. A coding agent reads the same YAML you do. If the YAML is silent on a condition, the agent has no signal either.
Descriptions and owners live in YAML, and only some owners are enforced
Per dbt's documentation, descriptions can be added to "models, columns, sources, and more", and they can include markdown. A check is simple: list every model and column in the exposed layer and count the ones with an empty description.
Owners are less tidy. Exposures take an owner block, and dbt's exposure page lists "owner: name or email required". For models, dbt's meta page says "meta accepts key-value pairs, is compiled into manifest.json", so an owner key on a model is a team convention that dbt will not enforce. Read manifest.json and flag every model whose meta has no owner.
- Documented: count models and columns with an empty description.
- Owner: count models whose meta has no owner key, and exposures with no owner block.
Tests, contracts and source freshness give the strongest native evidence
Per dbt's data tests page, "Data tests are assertions you make about your models and other resources in your dbt project". A passing test run is quality evidence produced by rule, where a description is only declared by a person. Check that each exposed model has tests on its keys and that the latest run is green.
Model contracts add a shape guarantee: "dbt will verify that your model's transformation will produce a dataset matching up with its contract, or it will fail to build." Check that contract enforced is true on the models an agent will read.
For freshness, dbt's source freshness page says the interface helps "determine if your source data freshness is meeting the service level agreement (SLA)". It adds that dbt build does not include freshness checks, so confirm a separate job runs them and run it at "at least double the frequency of your lowest SLA".
Metrics are defined in the semantic layer, and access is only between models
Per dbt's MetricFlow page, it "helps you define and manage the logic for your company's metrics". Check that every number an agent may report has a metric or semantic model behind it, so the agent queries a definition instead of writing its own SQL.
Access is the partial case. dbt's model access page describes private models, and its group page says a grouped resource can "reference private models within the same group". That controls which models may reference which. It does not say whether an AI agent may read a table, so mark the access condition as set by a person.
The ten conditions on dbt, with the gap for each
Each row names the dbt feature and the doc it comes from. Gaps are stated as none found on 13 Nov 2026 where the pages read showed nothing.
| Condition | What the platform gives you | Gap |
|---|---|---|
| Named owner | Exposure owner field; meta key by convention (dbt meta and exposures docs) | Models carry no enforced owner |
| Documented | description on models, columns, sources (dbt description and documentation docs) | No score for coverage; you count it |
| Certified | none found on 13 Nov 2026 | No certified or endorsed status |
| Quality tests pass | Data tests and unit tests (dbt data tests docs) | Pass or fail only for tests someone wrote |
| Fresh | Source freshness checks (dbt source freshness docs) | Sources only, not every model |
| Sensitivity classified | none found on 13 Nov 2026 | meta can hold a tag, dbt does not interpret it |
| Lineage known | Exposures list the resources that feed them (dbt exposures docs) | Downstream use outside dbt is declared by hand |
| Active use | none found on 13 Nov 2026 | Exposure maturity is typed in by a person |
| Metric defined | MetricFlow metrics and semantic models (dbt MetricFlow docs) | Covers only what you modeled |
| Access policy allows AI use | Model access and groups (dbt model access docs) | Controls model references, not agent reads |
What dbt leaves to another tool
Certification, classification of sensitive columns and real usage counts need something outside dbt. dbt's exposure page notes that maturity such as "high" is declared in YAML, which is a claim, not a measurement. Of the 17 tools ranked on this site, the ones listed below come from evidence that names dbt.
Lineage as a subject and sensitive-data exposure are covered elsewhere.
15 of 17 ranked tools name dbt in their evidence
This table is built from the systems each tool names, read 13 Nov 2026. A tool that names dbt can label its assets only as far as its own documentation shows; the marks come from the Readiness Coverage Matrix.
| Rank | Tool | Named as | Reach | Conditions by rule |
|---|---|---|---|---|
| 1 | OvalEdge | dbt Cloud, dbt Core | 5 | 6 of 10 |
| 2 | Euno | dbt Core, dbt Cloud, dbt Projects on Snowflake | 6 | 8 of 10 |
| 3 | DataHub | dbt | 5 | 6 of 10 |
| 4 | OpenMetadata | dbt, dbt Cloud | 6 | 7 of 10 |
| 5 | Atlan | dbt | 6 | 8 of 10 |
| 6 | Secoda | dbt Cloud, dbt Core | 5 | 8 of 10 |
| 7 | Databricks Unity Catalog | dbt (dbt-databricks adapter) | 5 | 7 of 10 |
| 8 | dbt | dbt Core, dbt Fusion, dbt platform | 5 | 6 of 10 |
| 9 | Alation | dbt | 5 | 7 of 10 |
| 10 | Ataccama | dbt | 4 | 5 of 10 |
| 12 | Monte Carlo | dbt Cloud, dbt Core | 4 | 5 of 10 |
| 13 | Snowflake Horizon | dbt Cloud, dbt Core | 3 | 6 of 10 |
| 14 | Precisely | dbt Cloud, dbt Core | 3 | 5 of 10 |
| 15 | Actian | dbt, dbt Cloud | 2 | 6 of 10 |
| 17 | Collibra | dbt | 3 | 5 of 10 |
Sources
- dbt Docs: description property, read 13 Nov 2026
- dbt Docs: meta config, read 13 Nov 2026
- dbt Docs: Add data tests to your DAG, read 13 Nov 2026
- dbt Docs: Source freshness, read 13 Nov 2026
- dbt Docs: Model contracts, read 13 Nov 2026
- dbt Docs: Add exposures to your DAG, read 13 Nov 2026
- dbt Docs: About MetricFlow, read 13 Nov 2026
- dbt Docs: Model access, read 13 Nov 2026
- dbt Docs: group config, read 13 Nov 2026
- dbt Docs: Documentation, read 13 Nov 2026
Last reviewed 13 Nov 2026.