How to Deploy LangAlpha with Docker: Step-by-Step Guide

LangAlpha ships with a full docker-compose setup: a backend (Python agent core + server), a web frontend, and a sandbox for safe code execution, plus PostgreSQL and Redis for infrastructure. Here's how to go from zero to your own private AI investment analyst.

Step 1 — Requirements

Step 2 — Get the code and configure

Clone the repository and copy the environment template:

git clone https://github.com/ginlix-ai/LangAlpha.git
cd LangAlpha
cp .env.example .env   # fill in your API keys

You'll need API keys for the LLM provider you plan to use (OpenAI, Anthropic, or a local model via an OpenAI-compatible endpoint) and optionally for market data sources you want the agent to query.

Step 3 — Start the stack

The compose file spins up PostgreSQL and Redis via the infra profile, then starts backend and web:

docker compose up -d

The backend listens on port 8000 (configurable via BACKEND_PORT), and the web UI is served alongside it. Give it a moment to pull images and migrate the database.

Step 4 — Create a workspace and start researching

Open the web UI, create a workspace for a research goal (e.g. "energy sector rotation"), and let the agent interview you about your style and constraints. It will produce a first deliverable, save everything to the workspace, and — because research compounds — tomorrow you can pick up exactly where you left off. Add MCP servers for live market data and the agent will discover and use their tools on demand.

Review verdict

LangAlpha is one of the most thoughtful open-source finance agents we've tested. The persistent-workspace model genuinely changes how you interact with an AI analyst — no more losing context between sessions. Setup is straightforward for anyone comfortable with docker-compose, and the MCP integration means your data sources are never locked in. The main caveat: quality of insights depends heavily on the LLM and data sources you plug in, and heavy research sessions want the recommended 8 GB of RAM.