dbt
SQL-first data transformation framework with tests and docs, plus a dbt MCP server, agent skills and a Developer Agent for AI coding tools
Overview
dbt lets data teams transform data inside their warehouse with SQL (and some Python), the way software engineers build apps: modular models, tests, documentation and version control. Its homepage calls it "the open standard for modern, AI-ready data transformation." You can run the open-source dbt engine yourself (Apache-2.0; the main repo now holds dbt v2.0, a Rust rewrite) or use the hosted dbt platform. The AI features that put it on this list are the dbt MCP server, which lets Claude, ChatGPT and other AI tools query your project, metrics and job history (OAuth connection in beta, May 2026); open-source dbt Agent Skills that teach coding agents dbt best practices (GA); and the dbt Developer Agent (preview), which drafts coordinated SQL, YAML, test and docs changes from a plain-language request. dbt Labs started as Fishtown Analytics in 2016, raised a $222M Series D at a $4.2B valuation in 2022 and passed $100M ARR in early 2025. Its merger with Fivetran closed on June 1, 2026, and the site now says "Fivetran + dbt Labs is now Fivetran." Also from Fivetran: Fivetran.
The Verdict
Who Should Use dbt?
Best For
- Analytics engineers who model warehouse data in SQL
- Teams that want tested, documented, version-controlled pipelines
- Companies giving AI agents governed metrics and lineage
Not Ideal For
- Moving data between systems (pair it with an ingestion tool)
- Large analyst teams on a tight budget (per-seat cloud pricing)
What's Great
- Automation and ease of use score highly on G2
- SQL-first, with built-in tests and docs
- Large community and a free open-source engine
Watch Out For
- Learning curve for beginners and complex initial setup
- Reviewers report dependency issues and limited functionality
- The Fusion engine is source-available (ELv2), not open source
Pricing
View all features & details
Key Features
- SQL and Python models with tests, docs and lineage
- dbt MCP server (remote and local; Apache-2.0)
- dbt Agent Skills (GA) and Developer Agent (preview)
- dbt Copilot, with bring-your-own-key for Anthropic, OpenAI, Azure OpenAI
- Semantic Layer for shared metrics
- Fusion engine, with a Fusion migration skill (beta)
Company
- Founded 2016 as Fishtown Analytics by Tristan Handy, Drew Banin, Connor McArthur
- Series D $222M at $4.2B (Feb 2022), led by Altimeter
- Merged with Fivetran, closed Jun 1, 2026; Handy is President
Real-World Usage
How It Compares
| Feature | dbt | Fivetran |
|---|---|---|
| Main job | Transform data in the warehouse | Move data into the warehouse |
| Open source | Yes (engine, Apache-2.0) | No |
| Agent access | MCP server, Agent Skills | Agent Context MCP (preview) |
| Free option | Developer plan, open source | Free plan |
Sources: dbt · Pricing · What shipped (May 2026) · dbt MCP server · Series D (Feb 2022) · Merger completion (Jun 2026)
Is this your tool? Add a badge