offdata ai — agentic AI for data modelers and data engineers

dbt

AI dbt project generator

Describe your domain in plain English. Ship a complete dbt project — sources, staging, marts, schema tests, and seeds — for every major cloud warehouse.

Get startedBook a demo
dbt project structuretree
dbt/
├── dbt_project.yml
├── models/
│   ├── staging/
│   │   ├── _sources.yml
│   │   ├── stg_app_db__customer.sql
│   │   ├── stg_app_db__plan.sql
│   │   └── stg_app_db__usage_event.sql
│   └── marts/
│       ├── core/
│       │   ├── dim_customer.sql      # SCD2
│       │   ├── dim_plan.sql
│       │   ├── fct_usage.sql         # incremental
│       │   └── fct_invoice.sql       # incremental
│       └── _schema.yml               # unique / not_null / relationships
├── seeds/
│   ├── dim_customer.csv
│   └── dim_plan.csv
└── tests/
    └── schema.yml

OffDataAI generates a complete dbt project from a single conversation. You get dbt_project.yml, source YAML with freshness, naming-consistent staging models, mart models for dimensions (including SCD2) and facts (including incremental), schema tests covering uniqueness and referential integrity, and realistic seed CSVs. The project is ready for dbt build on the first run.

What OffDataAI generates

  • Sources, staging, and marts

    Source YAML with freshness rules, lightweight staging models (one per source table, explicit casts), and mart models for facts and dimensions.

  • SCD2 dimensions, incremental facts

    Dimensions marked SCD2 emit valid_from / valid_to logic and surrogate keys. Facts emit incremental materializations with the right unique_key.

  • Schema tests included

    Every project ships with unique, not_null, and relationships tests covering primary keys, foreign keys, and natural keys.

  • Seeds that respect foreign keys

    Realistic CSV seeds with referentially consistent values — prototype your warehouse without source connections.

Frequently asked questions

Can OffDataAI generate a complete dbt project from plain English?
Yes. From a paragraph describing your domain, OffDataAI generates a full dbt project: dbt_project.yml, source YAML, staging models, mart models (dimensions and facts), schema tests, and seed CSVs. dbt build works on the first run.
Which warehouses are supported in the generated dbt project?
All major warehouses: Snowflake, Google BigQuery, Databricks (Delta), Amazon Redshift, PostgreSQL, Microsoft Synapse, and Microsoft Fabric. The dbt models include platform-specific configs (cluster_by, partition_by, dist_key, etc.).
Does it generate SCD2 models?
Yes. For dimensions you mark as SCD2 in the conversation, OffDataAI emits valid_from / valid_to windows, surrogate keys via dbt_utils.generate_surrogate_key, and the right window functions for change tracking.
Does it generate schema tests?
Yes. Every generated project includes schema.yml with unique, not_null, and relationships tests covering primary keys, foreign keys, and natural keys.
Does it generate seed data?
Yes. OffDataAI generates realistic seed CSVs that respect foreign-key relationships, plausible distributions, and any SCD2 windowing rules.

Related generators and modeling guides

Your data warehouse is one conversation away.

Describe your domain, or open one of 150+ production-grade templates. ERDs, DDL, and a complete dbt project — generated in under a minute.

Get startedTalk to usBrowse templates