Skip to main content
DinoAI can analyze your existing models, understand how they relate, and create well-structured intermediate and marts models with the right joins. In this guide you generate a marts model that combines several base models.
Prerequisites
  • DinoAI copilot access in your Paradime workspace.
  • A dbt™ project with base or staging models to combine.
Estimated time: 5 minutes.

Steps

1

Add the base models as context

In DinoAI, click the ”@” symbol and select the relevant base models so DinoAI knows exactly which tables to include and how they relate.
Adding the relevant base models as context is the most reliable way to get correct join conditions.
2

Enter your prompt

Describe the model you want to build. For example:
I want to create an intermediate model that lets me understand data across races as well as constructor and driver standings for the season.
DinoAI analyzes the structure of the models you provide, identifies relationships between tables (for example, foreign keys), and creates a new model with well-structured CTEs and join conditions. If configured, it applies your .dinorules preferences.
3

Review the generated model

Review the SQL before accepting it. The output uses CTEs for each source and applies the join conditions DinoAI inferred from the relationships.
4

Accept and test the model

Accept the changes to create the file, then build the model to confirm the joins and business logic produce the expected results.
The model builds without errors and combines the source tables with the relationships you expected. If a join returns duplicate or missing rows, verify the join keys in the generated CTEs against your base models.

Next steps

Generate documentation

Add model and column descriptions with tests.

Configure data pipelines

Set materializations, schemas, and tags for your models.

DinoAI use cases

Explore more ways to work with DinoAI.

Copilot reference

How context and Agent Mode work.