Your dbt™ project's biggest problem isn't the SQL — it's the context that never makes it in. The spec doc in Drive. The "official" definitions in a Sheet. The data contract nobody remembers to enforce. DinoAI's new Google Workspace integrations pull that context directly into your development workflow, so you can stop translating and start shipping.

Three new tools: Google Docs, Google Drive Search, and Google Sheets. They're most useful when chained together with DinoAI's action tools (Terminal, File System, GitHub PR) inside a single Agent Mode workflow.

Workflow 1: Spec Doc → dbt™ Models and Tests

A product manager drops a Google Doc with requirements. Instead of reading it in one tab and building in another, point DinoAI at it directly.

Read this Google Doc: <https://docs.google.com/document/d/1abc...xyz>
Generate dbt source definitions, a staging model, a marts model,
and a schema.yml with column descriptions and tests.
Then run dbt compile to validate.

DinoAI extracts the business logic, scaffolds the assets, and validates them. If dbt compile surfaces ref errors, it fixes them before returning results to you.

Workflow 2: Data Contract → Enforced Source Definitions

Contracts live in Drive. Models get built without them. DinoAI can close that loop.

Search Google Drive for the latest 'orders data contract' modified this month.
Read it and generate a dbt source definition with column-level tests —
unique, not_null, accepted_values where specified.
Embed the contract doc URL in the meta block.

The modification date filter ensures you're working from the current version, not a stale copy. The embedded URL in meta makes the source definition traceable back to the document that mandated it. When the contract updates, re-run the same prompt.

Workflow 3: Google Sheet → dbt™ Seed

Reference data living in Sheets — territory mappings, cost center codes, fiscal calendars — should be in your dbt™ project as tested, documented seeds. Now it can be, in one prompt.

Read the Google Sheet 1uXabc... (sheet: territory_mapping).
Create seeds/territory_mapping.csv.
Generate a schema.yml with not_null tests on territory_id
and accepted_values for region based on what's in the sheet.
Run dbt seed and dbt test --select territory_mapping.

DinoAI infers types, maps distinct values to accepted_values tests automatically, writes the seed, and runs a full validation cycle. For multi-tab workbooks, specify the sheet name or GID — it removes ambiguity.

Workflow 4: Drive Content → dbt™ Documentation

Most undocumented models aren't undocumented because the knowledge doesn't exist — it's because nobody has time to find it and write YAML. DinoAI can do that automatically.

Search Drive for docs mentioning 'customer lifetime value' or 'LTV methodology'.
Read the most relevant one and generate column-level descriptions
for models/finance/fct_customer_ltv.sql.
Match the language in the doc — don't use generic descriptions.

The "match the language" instruction matters — it keeps descriptions grounded in your team's actual definitions rather than generic boilerplate. This scales: point DinoAI at a Drive folder and a models directory and let it run.

Workflow 5: Spec → Pull Request

The full loop — from Google Doc to committed, PR-ready code.

Read the spec at <https://docs.google.com/document/d/1xyz>...
Create a feature branch called feature/refunds-model.
Scaffold the refunds staging and marts models from the spec.
Run dbt build --select refunds+.
If tests pass, commit and open a PR linking back to the spec doc.

DinoAI chains Docs → File System → Terminal → GitHub PR in one workflow. The PR description includes a spec summary and a direct link to the source document — code and requirements stay connected.

Workflow 6: Multi-Sheet Dimensional Modeling

When business definitions are spread across multiple spreadsheets, DinoAI can read them all and reason across them.

Read three Google Sheets:
1. 1abc... (sheet: products) — product dimension definitions
2. 1def... (sheet: channels) — channel taxonomy
3. 1ghi... (sheet: fiscal_calendar) — fiscal week offsets

Generate dim_products, dim_channels, and a fiscal_calendar seed.
Create a schema.yml covering all three with appropriate tests.
Run dbt build --select dim_products dim_channels fiscal_calendar.

Multi-source reconciliation that normally takes an afternoon happens in a single prompt.

A Few Things Worth Knowing

Filter Drive searches by date. Always include a modification date in Drive Search prompts when looking for contracts or specs. Stale docs produce wrong output.

Anchor to source URLs. Ask DinoAI to embed doc URLs in meta blocks and PR descriptions. Traceability from dbt™ asset back to source document is the point.

Let DinoAI handle the debug loop. When a workflow includes dbt build or dbt test, DinoAI reads error output and iterates. Don't interrupt it at the first failure.

Keep sheets clean. Headers in row 1, consistent column types. The Sheets Tool performs significantly better with well-structured data.

All three tools require a connected Google account in Paradime. Setup: Google Docs · Google Drive · Google Sheets

Full tool docs: Google Docs Tool · Google Drive Search Tool · Google Sheets Tool