Sora

Vero

Vero is a modern, open-source analytics and AI orchestration stack — designed to help organizations work data-driven and agent-driven without relying on black-box SaaS platforms or heavyweight enterprise BI tools.

It unifies best-in-class open technologies into a modular architecture that enables you to ingest, model, visualize, and query data — via both interactive dashboards and natural language agents.

But Vero goes beyond analytics.

It introduces support for Agentic AI, enabling multi-agent workflows, dynamic orchestration, and performance-aware coordination — allowing AI systems to take real action in complex environments while remaining observable, auditable, and governed.

Why Vero?

Make it easy for small and mid-sized teams to adopt a modern, open, and self-hosted analytics + AI stack — without giving up control.

❌ Common Pain Points

• Opaque vendor platforms with unclear logic
• Complex, inflexible data tools that resist integration
• Data sent off to third-party clouds just to get basic metrics
• No insight into what your AI agents are doing — or why
• Difficulty scaling AI workflows beyond one-off scripts

✅ What Vero Solves

Vero was built to address these challenges with a clean, composable reference architecture that combines modern analytics with intelligent, governed agent orchestration.

🚀 Our Goals

• Agent-first orchestration: Native support for multi-agent environments with workflow automation, decision delegation, and performance-aware orchestration
• Self-hosted analytics: Run the entire stack on your own infra — no vendor lock-in, no hidden fees
• Data sovereignty: Your data stays where it belongs — inside your infrastructure
• Modular & modern: Built on today’s best open-source tools (Cube.js, Metabase, n8n, Streamlit, PostgreSQL, etc.)
• Natural language interfaces: Enable users to ask complex questions — and get answers — through integrated LLM agents
• Scalable architecture: Designed for environments that need both real-time insights and AI actionability
• Team-friendly: Built to support engineers, analysts, and domain experts equally — no specialization required

With Vero, you’re not just consuming data — you’re building intelligent systems that reason, decide, and act.

And you own the full pipeline.

From ingestion to visualization, from raw facts to orchestrated agents, Vero gives your team the clarity, control, and composability to move from insight to action — securely, transparently, and at scale.

🧱 Architecture Overview

Vero is made up of six layers:

1. Data Ingestion – powered by Airbyte
2. Data Warehouse – default: PostgreSQL (replaceable with Apache Doris or DuckDB)
3. Semantic Modeling – Cube.js
4. AI Agent Interface – Agno + MCP server
5. Agentic AI Workflow Orchestration – n8n for automating agentic tasks and follow-ups
6. BI & Dashboards – Metabase (optional: Superset, Tableau, Power BI, etc.)

Each layer is loosely coupled and can be replaced or extended as needed.

📦 What's Included?

• Prebuilt Docker environment for local or on-prem deployment
• Sample dataset based on the Contoso Retail model
• Ready-to-use semantic models and views in Cube.js
• Natural language query agent powered by GPT or Claude
• Automated workflows and agentic AI orchestration with n8n
• Traditional dashboards and visualizations via Metabase
• NGINX reverse proxy setup for routing and traffic management
• Let's Encrypt SSL certificates automated via nginx-proxy and docker-gen for secure HTTPS out of the box

🧪 Quickstart

# 1. Clone the repository
$ git clone https://github.com/kpi-zone/Vero.git
$ cd Vero

🐳 Choose the Right Docker Compose File

Vero supports two versions of the Docker Compose configuration.

These two versions are necessary because the default Metabase image (metabase/metabase:latest) does not support ARM64 architecture out-of-the-box. If you're running Vero on an ARM64 machine (like Apple Silicon), you'll need a compatible image to ensure Metabase starts correctly.

File	Description
docker-compose.yaml	Default version for most x86_64 environments
docker-compose.arm64.yaml	Special version for ARM-based systems like Apple Silicon (M1/M2 chips)

To check your system architecture:

uname -m

• x86_64 → Use docker-compose.yaml
• arm64 → Use docker-compose.arm64.yaml

Run the stack:

# For x86_64:
docker compose --env-file .env.dev -f docker-compose.yaml up --build

# For ARM64 (Apple Silicon):
docker compose --env-file .env.dev -f docker-compose.arm64.yaml up --build

See docs/quickstart.md for full instructions and environment setup.

📚 Documentation

• Architecture Overview
• Quickstart
• Docker & Runtime Config
• Environment Variables
• Semantic Modeling (Cube.js)
• BI Layer (Metabase)
• AI Agent Layer (MCP + Agno)
• Warehouse Layer (PostgreSQL)

❤️ Contributions Welcome

This is a community-friendly reference project. We welcome:

• Feature suggestions and bug reports
• Docs, UX, and onboarding improvements
• New data sources, transformations, and AI plugins

Feel free to open a PR or issue!

📄 License

Vero is licensed under the GPL-3.0 License. All included third-party projects follow their respective licenses.

For support, ideas, or integration help, reach out via GitHub Discussions or open an issue. We’re building Vero to be the foundation for modern, self-managed analytics — for everyone.