From business requirements to production-ready data architecture
and AI-ready enterprise context. Six stages, one platform, one model underneath them all, so nothing gets lost between the whiteboard, the warehouse and the agent.
01
Design
Logical, physical and dimensional models.
02
Generate
SQL, dbt, documentation and migrations.
03
Define
Business metrics, semantic models and ontology.
04
Connect
Real warehouse schemas and metadata.
05
Govern
Drift, lineage, impact analysis and approvals.
06
Activate
Make trusted context available to AI agents.
offdataai is not
- An AI ER diagram generator.
- Another dbt alternative.
- Another data catalog.
- Another ontology editor.
It's the whole path, in one platform.
Business requirements
conversations, briefs, templates
Production-ready data architecture
models, SQL, dbt, migrations
AI-ready enterprise context
metrics, ontology, governed access
Requirements in. Architecture and AI-ready context out.
Every stage builds on the one before it, in one place, so nothing gets lost between the whiteboard, the warehouse and the agent.
01
Design
Logical, physical and dimensional models.
Start from a conversation, a template or your own schema. Shape entities and relationships once, then view them as logical, physical or star-schema models.
- customer_idkey
- name
- segment
- loyalty_tier
- order_idkey
- customer_idkey
- product_idkey
- order_total
- product_idkey
- category
- price
- date_idkey
- month
- quarter
02
Generate
SQL, dbt, documentation and migrations.
Production-ready DDL for your warehouse, a dbt project, docs your team will actually read, and migrations every time the model changes.
create table analytics.fct_orders ( order_id varchar primary key, customer_id varchar references dim_customer, order_total number(12,2) not null, ordered_at timestamp_ntz);
03
Define
Business metrics, semantic models and ontology.
Agree what “net revenue” and “active customer” mean, once. Metrics, semantic models and an ontology sit right on top of the physical model.
Metric
Net revenue
sum(order_total) − refunds
Metric
Active customer
ordered in the last 90 days
Semantic model · ontology · one shared vocabulary
04
Connect
Real warehouse schemas and metadata.
Point it at Snowflake, BigQuery, Databricks, Postgres and more. Real tables, columns and usage flow in, so the design reflects what's actually deployed.
Your model
412 tables
6,180 columns · usage · owners
Synced 2 minutes ago
05
Govern
Drift, lineage, impact analysis and approvals.
See when production drifts from the design, what a change would break downstream, and who signed it off, before it ships.
dim_customer · drift detected
2 changesImpact
3 dbt models · 2 dashboards · 1 metric
Approval
✓ Approved by data council
06
Activate
Make trusted context available to AI agents.
Your models, metrics, definitions and access rules become context AI agents can use over MCP, so their answers are grounded in what your business agreed.
$4.82M, up 12% on the quarter before. Loyalty customers were 38% of net revenue.
select sum(order_total) - sum(refunds) from fct_orders join dim_customer using (customer_id) where loyalty_tier is not null and quarter = '2026-Q3'
✓ Answer grounded in your agreed definitions
Change, made safe.
Production drifts, models evolve and agents depend on both. Governance is built into the same model, not bolted on afterwards.
Drift
Snapshot a connected database and see exactly what changed since: tables and columns added, dropped or retyped, against the model you designed.
Migrations
Every edit is a new version of the model. Diff any two and get a reviewable ALTER/CREATE script for your warehouse.
Impact and lineage
The catalog shows which concepts and metrics use each table and column, so you know what a change touches before it ships.
Classification and masking
Tag PII and sensitivity per column, see masking coverage, and generate platform-native masking policies.
Review and approvals
Viewer and editor roles, comments on the model, and changes pushed to GitHub as pull requests your team approves.
Explore each stage
From requirements to AI-ready context, in one platform.
Describe your business in a conversation, start from a template or connect your own database. Your first model is ready in minutes.