offdataai
Database Modeler

Reverse-engineer any database into a clean data model

Connect Postgres, MySQL, Snowflake, BigQuery, Redshift or Databricks. offdataai reads the catalog, finds the relationships nobody declared, classifies every table, and builds a project you can modernize.

  • From $1/month
  • 8 SQL targets
  • 174 templates
Connected · PostgreSQLMetadata only
PostgreSQLSnowflakeDatabricksrows read: 0
sales.orders14 cols
sales.order_items9 cols
crm.accounts22 cols
catalog.products17 cols
ref.currency4 cols

Suggested relationships

order_items.order_id
→ orders.order_id
98% confidentreview
orders.cust_ref
→ accounts.account_id
91% confidentreview
Mirror 1:1Remodel as star
Reads tables, columns, keys and comments, never rows, and suggests the joins nobody declared.
Watch the demo

See it in action

A narrated walkthrough, recorded in the real app.

How it works

Three steps, minutes not months

1

Describe or import

Answer a short interview, open a template, draw on the canvas or connect a live database.

2

Review a validated model

Keys, grain, history and PII are checked before any code is written. Refine it in plain English.

3

Ship and keep it current

Native DDL, dbt, seeds, docs and an ontology. Push to Git, diff versions, generate migrations.

Overview

Why teams choose offdataai

app.offdataai.com
A live database read into the catalog, tables classified and keys suggested

A live database read into the catalog, tables classified and keys suggested

  1. 01

    Most real databases were never designed on a whiteboard.

  2. 02

    Foreign keys are missing, staging tables sit next to facts, and the only documentation is in someone's head.

  3. 03

    Rebuilding a model from that by hand takes weeks.

  4. 04

    The Database Modeler does the archaeology for you.

  5. 05

    It reads only the catalog, never your rows, then infers undeclared relationships with a confidence score and the evidence behind each one.

  6. 06

    Every table is classified as a fact, dimension, bridge, lookup, staging table and more, with grain, PII and history detected.

  7. 07

    Review the result, then build a project: mirror the schema 1:1, or let AI remodel it as a star schema or Data Vault for the warehouse you are moving to.

Real output

The code it writes, not a sketch

This is the shape of what offdataai generates here, straight from the validated model. Copy it, download it, or push it to Git.

  • Catalog-only introspection
  • Relationship inference you can review
  • Automatic classification
database-modeler · analysissnowflake
1source snowflake · ANALYTICS.PUBLIC
2tables 142 read (catalog only, no rows)
3
4inferred relationships
5 orders.cust_id → customers.id 0.94 ✓
6 order_items.order_id → orders.id 0.97 ✓
7 payments.ord_ref → orders.id 0.71 review
8
9classification
10 orders fact grain: one row per order
11 customers dimension SCD 2 · PII: email, phone
12 order_items fact grain: one row per line
13 stg_orders_raw staging excluded
14
15build remodel → Kimball · target: snowflake
Capabilities

What the Database Modeler does

  • Catalog-only introspection

    Tables, views, columns, keys, indexes, constraints and comments, read from the catalog. No rows are selected, and credentials are encrypted at rest.

  • Relationship inference you can review

    Undeclared foreign keys are suggested with a confidence score and evidence. Accept, reject or bulk-accept; only accepted ones are used.

  • Automatic classification

    Facts, dimensions, bridges, lookups, staging and audit tables, plus column roles, PII, SCD type, grain and subject area.

  • Mirror or remodel

    Keep the schema 1:1 for documentation, or remodel it as a Kimball star schema or Data Vault 2.0 for a new platform.

  • Schema drift

    Re-analyze a source and compare snapshots to see exactly what changed before you rebuild.

  • Write back to Unity Catalog

    On Databricks, learn joins from query history, apply the keys, descriptions and PII tags you approved, and publish a metric view so Genie stops double-counting.

  • From database to ontology

    Use an imported schema as an ontology source and merge it with your warehouse model into one business vocabulary.

0

Validated templates

0

SQL targets

0

Modeling paradigms

0+

Industries covered

Generates native SQL and dbt for 8 platforms · connects live to Snowflake, Databricks, BigQuery, Redshift and Postgres

SnowflakeBigQueryDatabricksRedshiftPostgreSQLSynapseFabricDuckDBdbtSnowflakeBigQueryDatabricksRedshiftPostgreSQLSynapseFabricDuckDBdbt
FAQ

Frequently asked questions

Which databases can the Database Modeler read?
PostgreSQL, MySQL, Snowflake, Google BigQuery, Amazon Redshift and Databricks (Unity Catalog).
Does offdataai read my data?
No. The Database Modeler reads only the database catalog: tables, views, columns and types, declared primary and foreign keys, indexes, constraints, comments and estimated row counts. It never selects rows from your tables, so no customer data is read or sent to an AI model. Connection credentials are encrypted at rest. Two opt-in Databricks features go further: learning joins reads query text from system.query.history and keeps only the join columns and counts, and publishing a metric view queries each measure once so you can check the totals.
What if my database has no foreign keys declared?
That is common, so offdataai infers them. It looks at column naming, type compatibility, whether the target is a key, indexes, joins inside views and NOT NULL constraints, and gives each suggested relationship a confidence score and its evidence. You accept or reject them, or bulk-accept the high-confidence ones.
How are tables classified?
A rule-based pass always runs and labels tables as fact, dimension, bridge, lookup, operational, staging, audit or junk, and columns by role such as key, measure or flag. It also detects PII, SCD type, grain and subject area. An optional AI pass refines the tables you select. You can edit any classification.
What can I build from an imported schema?
A full offdataai project. Mirror keeps your tables 1:1 and is free of AI credits; Remodel redesigns the schema as a Kimball star schema or Data Vault 2.0 using AI. Either way you choose a target platform (PostgreSQL, Snowflake, BigQuery, Databricks, Redshift, Synapse or Fabric) and get the ERD, DDL and dbt project. The imported schema can also feed an ontology.
Can it detect schema changes?
Yes. Re-analyze a connection and offdataai shows the drift between snapshots, so you can see what changed in the source before rebuilding.
What does it add for Databricks?
Three things. With “Learn joins from query history” on, it reads the last 90 days of system.query.history to find the joins your team actually runs, and keeps only the join columns and counts. With “Allow write-back to Unity Catalog” on, Apply to Unity Catalog writes the keys, descriptions and PII tags you approved to your tables; each change is a statement you can read first, and tags are only added, never changed when Unity Catalog already holds a different value. And it can publish a Unity Catalog metric view from the reviewed model, so Genie and BI tools sum each measure at its own grain instead of double-counting header amounts after a join.
Which plan includes the Database Modeler?
The Database Modeler is part of the Pro plan, along with live connections, governance, the public API, the MCP server and publishing to Git. Team adds an organization workspace on top. AI steps such as remodeling and the AI classification pass use credits; a 1:1 mirror does not.

Part of the offdataai platform

The business context behind accurate AI answers, built from the data model up.

See all six stages →
  1. 01DesignLogical, physical and dimensional models.
  2. 02GenerateSQL, dbt, documentation and migrations.
  3. 03DefineBusiness metrics, semantic models and ontology.
  4. 04ConnectReal warehouse schemas and metadata.
  5. 05GovernDrift, impact analysis, tests and approvals.
  6. 06ActivateMake trusted context available to AI agents.

This page covers Design and Connect. Every stage works on the same model, so what you design here carries through to the rest.

Accuracy starts with context

Give your AI agents context they can trust.

Design the model, define what your metrics mean, and serve that context to every agent and BI tool, on the warehouse you already run.

Book a demoGet started

Self-serve plans available