How to Deploy Verba with Docker: Step-by-Step Setup Guide

This guide walks you through getting a production-ready Verba RAG system running with Docker.

Prerequisites

Docker and Docker Compose installed
At least 4GB of RAM recommended
An API key for your LLM provider (OpenAI, Cohere) or Ollama for local operation

Step 1: Pull the Image

docker pull semitechnologies/verba:latest

Step 2: Create docker-compose.yml

version: "3.8"
services:
  weaviate:
    image: semitechnologies/weaviate:latest
    command: --host 0.0.0.0 --port 8080 --scheme http
    environment:
      OPENAI_APIKEY: ${OPENAI_API_KEY}
      QUERY_DEFAULTS_LIMIT: 25
      AUTHENTICATION_ANONYMOUS_ACCESS_ENABLED: "true"
      PERSISTENCE_DATA_PATH: /var/lib/weaviate
    volumes:
      - ./data/weaviate:/var/lib/weaviate
    restart: unless-stopped

  verba:
    image: semitechnologies/verba:latest
    ports:
      - "8081:8080"
    environment:
      OPENAI_API_KEY: ${OPENAI_API_KEY}
      WEAVIATE_URL_VERBA: http://weaviate:8080
    volumes:
      - ./data/verba:/data
    restart: unless-stopped
    depends_on:
      - weaviate

Step 3: Launch and Access

OPENAI_API_KEY=sk-your-key docker compose up -d

For Ollama (local): docker compose -f docker-compose.ollama.yml up -d

Open http://localhost:8081 in your browser. Upload documents, choose your embedder, and start asking questions.

How It Works

When you upload a document, Verba chunks the text, generates embeddings via your configured embedder, and stores them in Weaviate. When you ask a question, it finds the most relevant chunks, sends them to the LLM along with your question, and returns a cited answer. All processing happens on your own hardware.

Troubleshooting Tips

Connection refused: Weaviate may not be ready yet — check docker compose logs weaviate
No results: Verify your document shows in the data explorer tab
API errors: Double-check your LLM API key

Next Steps

Once running, explore different embedders, adjust chunk sizes, or connect external data sources. Verba's modular design makes experimentation easy.