Maxa

Maxa provides an agentic platform that generates and maintains multi-system data pipelines for finance and operations use cases. Its workflow has three stages: agents ingest unstructured business context — meeting transcripts, dashboard screenshots, spreadsheets, PDFs, sample data, ERP configs, DDL/schemas, existing dbt projects and BI models — and convert it into a unified specification, surfacing gaps for stakeholders to resolve before code is written. In the second stage, "Maxa Autopilot" writes dbt code while continuously validating it against the spec, historical reports and known business facts, with agents cross-checking each other's work until output figures match. In the third stage, the customer downloads the resulting dbt project and deploys it themselves.

The generated output is a standard dbt project containing star schemas and marts at arbitrary granularity, with schema, business-rule and reconciliation tests, documentation and end-to-end lineage traceable from source row to business requirement. Maxa states the code has no proprietary runtime dependency, lives in the customer's repository and continues to work if the customer changes warehouse, dbt tooling or vendor. The platform targets deployment on Databricks, Microsoft Fabric and Snowflake, and the resulting marts are intended to serve BI tools such as Tableau, Power BI, Looker and Sigma without an additional semantic modeling layer. Maxa also positions the models as a foundation for AI agents, embedding business context, lineage and metric definitions directly in the model rather than in separate semantic layers or ontologies.

On data handling, Maxa states it does not connect to or query customer production warehouses. Customers export schemas and a sample of rows — up to 10,000 per table — into a local database within their Maxa environment, where all agent work runs read-only against that copy.

Business model

Maxa sells access to an agentic pipeline-building platform; the site offers a free trial and a "talk to an expert" sales motion. The generated dbt code is owned and executed by the customer in their own repository and warehouse rather than on Maxa infrastructure.