Quick Start

Time: 5 minutes Prerequisites: Docker and Docker Compose installed

This guide gets DecisionBox running locally with Docker Compose. By the end, you'll have the dashboard open and ready to create your first project.

1. Clone and Start

git clone https://github.com/decisionbox-io/decisionbox-platform.git
cd decisionbox-platform

# Start all services
docker compose up -d

This starts four containers:

Service	Port	Description
MongoDB	27017	Database (projects, discoveries, secrets)
Qdrant	6333/6334	Vector store (schema-index + insight/recommendation vectors)
API	8080	REST API (not exposed publicly — dashboard proxies to it)
Dashboard	3000	Web UI

Wait about 10 seconds for all services to start, then open http://localhost:3000.

2. Create a Project

Click New Project and fill in:

1. Basics — Project name, domain (e.g., Gaming or Social Network), category (e.g., Match-3, Idle, Casual, Content Sharing)
2. Data Warehouse — Select your warehouse provider and enter connection details
3. AI Provider + Embedding — Select your LLM provider, model, and API key (Claude, OpenAI); also pick an embedding provider (required — schema indexing and /ask both depend on it). Providers that use cloud auth (Vertex AI, Bedrock, Ollama) don't need a key.

Click Create Project. Your API key is encrypted and stored per-project.

3. Verify Connection

Go to Settings → Data Warehouse or AI Provider tab and click Test Connection to verify your setup works before running a discovery.

4. Run Your First Discovery

Wait for Settings → Schema Index to reach Ready — the discovery button stays disabled until the schema index is built. Then click the Run discovery button in the top bar. The AI agent will:

1. Discover your warehouse table schemas
2. Autonomously write and execute SQL queries
3. Analyze results to find patterns
4. Validate findings against your data
5. Generate actionable recommendations

You can watch the progress live — each step shows what the AI is thinking, what SQL it wrote, and what it found.

5. Review Results

When the run completes, click the discovery card to see:

• Insights — Severity-ranked findings with confidence scores
• Recommendations — Specific action steps with impact estimates
• Transparency — Every SQL query the AI ran, its reasoning, and validation results

Next Steps

• Installation Guide — Other ways to install (binary, from source)
• Your First Discovery — Detailed walkthrough with explanations
• Configuring LLM Providers — Set up Claude, OpenAI, Vertex AI, Bedrock, or Ollama
• Configuring Warehouses — BigQuery or Redshift setup details

Stopping

# Stop all services (data preserved)
docker compose down

# Stop and remove all data
docker compose down -v