Overview
What is Vero?
Vero is a modern, open-source analytics stack that integrates an MCP (Model Context Protocol) server to enable natural language querying over your data. It is designed for small to mid-sized organizations that want to own their full analytics pipeline, from ingestion to insight, without relying on third-party SaaS platforms. The stack includes data ingestion (Airbyte), a warehouse (PostgreSQL by default), semantic modeling (Cube.js), an AI agent interface (Agno + MCP server), and BI dashboards (Metabase).
How to use Vero?
Clone the repository, choose the appropriate Docker Compose file (x86_64 or ARM64), set environment variables, and run docker compose up --build. The MCP server is part of the AI Agent layer and is configured as part of the stack.
Key features of Vero
- Open-source, vendor-neutral analytics stack
- Self-hosted with Docker for local or on-prem deployment
- Modular architecture with replaceable layers
- Natural language query via MCP server and AI agent
- Traditional dashboards with Metabase (or alternative BI tools)
- Sample Contoso Retail dataset included
Use cases of Vero
- Self-hosted business intelligence without external data exposure
- Natural language querying of sales and operational data
- Replacing costly SaaS analytics platforms with an open alternative
- Enabling data-driven decision-making for non-technical stakeholders
FAQ from Vero
What dependencies does Vero require?
Vero runs on Docker and includes PostgreSQL, Cube.js, Agno, Metabase, and an MCP server. It is designed to be self-contained with prebuilt Docker environments.
Where does Vero store data?
All data remains on your own servers. The default warehouse is PostgreSQL, and you can replace it with Apache Doris or DuckDB.
Does Vero require an external AI API?
Yes, the natural language agent uses GPT or Claude, so an API key for those services may be needed.
What transport or authentication does the MCP server use?
This is not specified in the provided README. Refer to the docs/ai-agent/mcp-server.md for details.
Can I replace the BI tool?
Yes, the BI layer is optional and can be replaced with Superset, Tableau, or Power BI.
