The job dbt does, without the stack it needs

dbt solved a real problem. But it was never meant to be the whole answer.

February 2026 · 4 min read · Perspective

The problem dbt solved

Raw data is messy. Every company knows this. Customer records live in the CRM, financial data in the ERP, marketing metrics across half a dozen platforms. Before any of it becomes useful, someone has to clean it, join it, and turn it into something the business can actually trust.

That's the transformation layer. For years, it was held together by scripts, spreadsheets, and tribal knowledge. dbt changed that. It gave data teams a proper framework: version-controlled SQL, dependency graphs, tests, documentation. Business logic finally lived in code, not in someone's head. That was a real step forward.

But dbt is one piece of many

dbt doesn't ingest data. You need Fivetran or Airbyte for that. It doesn't store it. You need Snowflake or BigQuery. And it doesn't help anyone outside the data team ask questions. You need Power BI, Tableau, or Looker for that. Multiple tools, multiple vendors, multiple contracts. And a data team to glue it all together.

For a mid-sized company, the total cost of this stack runs between $270k and $950k per year, most of it in people. The backlog never shrinks. Business users wait days or weeks for answers. And the people who actually understand the business, the ones who know what "revenue by cohort" should mean, are the furthest from the data.

What changes when AI handles the transformation layer

Ronja's agents write the same kind of transformation code that a data engineer would. Deterministic, versioned, traceable. But they do it on demand, triggered by the questions your team actually asks. When someone asks "what's our gross margin by product line?" and the data model doesn't exist yet, agents build it. Real code. Real DAGs. Auditable and reproducible.

The difference: business logic persists as infrastructure, not as a one-time chat response. Every model the agents build becomes part of your data layer. Tested, documented, and reusable. The platform gets smarter the more your team uses it, learning your terminology, your business rules, your metric definitions. Same rigor as dbt. No human operator maintaining every model.

One platform instead of four tools and a team

Ronja replaces the entire pre-AI data stack. Agents handle every layer, from ingestion to insight.

Ingestion: connect to 100+ sources, no Fivetran needed
Storage: built-in data lake, no Snowflake needed
Transformation: agents write versioned code, no dbt needed
Answers: dashboards, reports, and plain-language exploration

The job dbt does, without the stack it needs

The problem dbt solved

But dbt is one piece of many

What changes when AI handles the transformation layer

One platform instead of four tools and a team

See what your data stack looks like without the stack