The dbt Semantic Layer and offdataai — better together
dbt is where transformations and metrics-as-code live. offdataai works upstream: it designs the data model and what its metrics mean, validates them, then hands dbt a full project and an Apache Ossie semantic model.
- From $1/month
- 8 SQL targets
- 174 templates
1my_dbt_project/2 dbt_project.yml3 models/4 staging/ stg_orders.sql stg_customers.sql ...5 marts/ fct_orders.sql dim_customer.sql ...6 seeds/7 OSI/8 sales_ontology.json # Apache Ossie 0.1.1, exported by offdataai9 # datasets · fields · relationships · metrics1011$ dbt parse # dbt Core 1.12 reads OSI/ alongside your YAML
dbt Semantic Layer and offdataai, compared
| dbt Semantic Layer | offdataai | |
|---|---|---|
| What it is | Metrics defined in your dbt project with MetricFlow, on top of your dbt models | The data model, ontology and metrics, designed and validated before and alongside the dbt project |
| Authoring | YAML in the dbt project: semantic models, entities, dimensions and metrics | An AI interview, 174 templates or the Model Builder canvas for the model; an ontology of concepts, relations and metrics with team-scoped meanings and rulings |
| The dbt project itself | You write the models, sources and tests | Generates a full dbt project from the model, publishes it to GitHub, GitLab or Bitbucket, and imports existing dbt YAML |
| Querying metrics | Semantic Layer APIs and integrations, on dbt platform Starter or Enterprise tiers | MCP server and REST API with scoped keys; query execution on Postgres, Redshift and Databricks |
| AI agents | The dbt MCP server connects AI tools to the Semantic Layer | MCP server with 17 tools (concept search, join paths, ask_ontology, run_query), under row filters and PII masking, with the SQL each agent ran |
| Platforms | MetricFlow supports Snowflake, BigQuery, Databricks and Redshift, plus Postgres on dbt v1 | DDL for Snowflake, Databricks, BigQuery, Redshift, Postgres, Synapse, Microsoft Fabric and DuckDB |
| Open formats | MetricFlow is Apache 2.0; dbt Core 1.12 parses Apache Ossie documents (beta) | Exports and imports Apache Ossie (spec 0.1.1), plus OWL and JSON-LD |
Which one fits your team
Choose dbt Semantic Layer when
- Your analytics engineers already own the dbt project and want metric definitions in the same YAML, reviewed in the same pull requests.
- You query metrics from the BI tools and applications that integrate with the dbt Semantic Layer APIs.
- You are on a dbt platform Starter or Enterprise plan and want caching and exports managed there.
Choose offdataai when
- The model doesn't exist yet, or needs a redesign, and you want it designed from the business with the meaning captured at the same time.
- Several teams define the same term differently and you need a recorded ruling, a release process and approval before agents use a definition.
- Agents need the context directly over MCP, under row filters and PII masking, with a log of the SQL they ran.
Design and define in offdataai, build and run in dbt
offdataai and dbt meet in open files. offdataai generates the dbt project from a validated model and exports the ontology as an Apache Ossie semantic model, which dbt Core 1.12 reads from an OSI/ directory in the project (beta). Teams with an existing project can import its dbt YAML instead.
- 01
Design the model in offdataai, or import your dbt YAML.
- 02
Generate the dbt project and publish it to GitHub, GitLab or Bitbucket.
- 03
Define and release metrics in the ontology, with rulings where teams disagree.
- 04
Export Apache Ossie JSON into the project's OSI/ directory for dbt to parse.
See it in action
A narrated walkthrough, recorded in the real app.
Three steps, minutes not months
Describe or import
Answer a short interview, open a template, draw on the canvas or connect a live database.
Review a validated model
Keys, grain, history and PII are checked before any code is written. Refine it in plain English.
Ship and keep it current
Native DDL, dbt, seeds, docs and an ontology. Push to Git, diff versions, generate migrations.
Why teams choose offdataai

A validated ERD: keys, SCD2 history and PII flags on every table
- 01
The dbt Semantic Layer made metrics-as-code mainstream: define revenue once in the project, and every downstream tool gets the same number.
- 02
It assumes the dbt models already exist and that the people writing
YAMLknow what each team means by each term. - 03
offdataai handles what comes before and around that.
- 04
It designs the model from the business, captures each team's meaning with a ruling when they conflict, validates it, and generates the dbt project.
- 05
The definitions travel to dbt as an open Apache Ossie semantic model and to agents over MCP, so dbt stays the place you build and run.
What offdataai hands to dbt
A full dbt project
Sources, staging and mart models, tests and seeds, generated from the validated model.
Git publish
Push the project to GitHub, GitLab or Bitbucket for review in your normal workflow.
Migrations
Each model change becomes a new version, with the migration between versions generated.
Apache Ossie export
The ontology as an open semantic model, in a spec version dbt Core 1.12 accepts.
dbt YAML import
Start from an existing project instead of from scratch.
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
FabricDuckDB
FabricDuckDBFrequently asked questions
Does offdataai replace the dbt Semantic Layer?
How does an offdataai ontology get into dbt?
Can offdataai generate the dbt project too?
Can I bring an existing dbt project into offdataai?
What does offdataai add on top of metrics-as-code?
More comparisons
- Atlan vs offdataaiAtlan builds context from the assets you already have. offdataai captures meaning when the model is designed, validates it and serves it to agents.
- SqlDBM vs offdataaiBoth are cloud modelers. offdataai starts from a conversation or a template and treats meaning, validation and agent access as the core, in open formats.
- Databricks Unity Catalog vs offdataaiUnity Catalog metric views are a strong home for meaning inside Databricks. offdataai keeps the definitions platform-neutral and publishes them to Databricks too.
- Power BI semantic models vs offdataaiPower BI semantic models serve Power BI reports well. offdataai keeps business definitions outside any one BI tool, for every agent and warehouse.
- All comparisonsCatalogs, modelers, warehouse semantics and BI models, side by side.
- dbt project generatorA complete dbt project from a validated model.
Part of the offdataai platform
The business context behind accurate AI answers, built from the data model up.
- 01DesignLogical, physical and dimensional models.
- 02GenerateSQL, dbt, documentation and migrations.
- 03DefineBusiness metrics, semantic models and ontology.
- 04ConnectReal warehouse schemas and metadata.
- 05GovernDrift, impact analysis, tests and approvals.
- 06ActivateMake trusted context available to AI agents.
This page covers Generate and Connect. Every stage works on the same model, so what you design here carries through to the rest.
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.
dbt, dbt Core and MetricFlow are trademarks of dbt Labs, Inc. offdataai is not affiliated with or endorsed by dbt Labs. Comparison based on dbt's public documentation as of November 2026.
