The Scenario
You are an MLOps engineer. The data science lead has asked for a model lifecycle dashboard before the quarterly review on Monday. She wants every registered model in the Databricks Unity Catalog model registry — model name, catalog, schema, latest version number, and status — in a Google Sheet so the team can see what is in production, what is staging, and what needs to be retired.
You have thirty-seven registered models across three catalogs. The Unity Catalog UI shows them one at a time.
The bad version of the weekend before the review:
- You open the Unity Catalog UI and navigate to the model registry
- You click each model to find its latest version and status
- You copy the details into a spreadsheet by hand
- You realize you missed the catalog name on half the rows and have to go back
- You spend Sunday evening on a task that should have taken fifteen minutes
- You show up Monday with a sheet that already has a stale version status in it.
The fast version is one prompt Friday afternoon.
The Easy Way: One Prompt in SheetXAI
SheetXAI is an AI agent inside your Google Sheet that reads your Databricks Unity Catalog model registry directly, so you do not have to click through the UI or write Python to call the API.
Open the SheetXAI sidebar and type:
Export all registered models from my Databricks Unity Catalog model registry. Write model name, catalog, schema, latest version number, and status into this sheet with headers in row 1. Sort by catalog name, then model name.
SheetXAI calls the Unity Catalog model registry APIs, pages through all registered models, and writes the full inventory into the sheet. Thirty-seven rows, one prompt.
What You Get
A Google Sheet with the full model registry inventory:
- Column A — model name
- Column B — catalog
- Column C — schema
- Column D — latest version number
- Column E — status (Production, Staging, Archived, None)
Ready for the data science lead to annotate. She can add a "Next Action" column and sort by status to see what needs to be retired versus what is safe to promote.
What If the Data Is Not Quite Ready
Model registry inventories often need more than just the list. SheetXAI handles the follow-on questions in the same prompt.
When the status field uses non-standard values
Some models were registered before the team standardized on Production, Staging, and Archived as the canonical statuses. Old records say things like "PROD" or "prod" or "staging-v2."
Export all registered models from my Databricks Unity Catalog model registry with model name, catalog, schema, latest version number, and status. In a new column called "Normalized Status," apply this mapping: PROD, prod, Production → Production; staging, staging-v2, Staging → Staging; archived, ARCHIVE → Archived. Write "UNKNOWN" for anything else. Sort by Normalized Status.
When the data science lead wants version count alongside latest version
She wants to see how many versions each model has accumulated — a sign of active development versus a stale model nobody is updating.
Export all registered model versions from my Databricks Unity Catalog model registry. For each model, count the total number of versions and write model name, catalog, latest version number, total version count, and status into this sheet. Sort by total version count descending.
When only production-status models matter for the Monday review
The quarterly review is about what is live. Staging and archived can wait.
Export all registered models from my Databricks Unity Catalog where the latest version status is Production. Write model name, catalog, schema, latest version number, and the run ID of the latest version into this sheet. Sort by model name.
When the dashboard needs the model inventory plus recent training run metrics in one view
The data science lead wants to see not just which models exist but what their latest training run performance looked like — loss, accuracy, or whatever metric the team tracks.
Export all registered models from my Databricks Unity Catalog. For each model, write model name, catalog, schema, and latest version number into columns A through D. Then for each model's latest version, look up the associated MLflow run and write the run's logged metric called "eval_accuracy" into column E. If no eval_accuracy metric exists, write "N/A." Sort by eval_accuracy descending so the best-performing models appear first.
The pattern: the model governance artifact that always gets deferred to "someday" takes one prompt on a Friday afternoon.
Try It
Get the 7-day free trial of SheetXAI and open a blank Google Sheet, then ask it to pull your full Databricks model registry inventory. The Databricks integration is included in every plan. For related workflows, see how to export a Unity Catalog table inventory or the Databricks in Google Sheets overview.