Deploy and Host Bytebot Protected AI Desktop Stack on Railway

AI desktop agent with Postgres and a generated-password access gateway.

About Hosting Bytebot Protected AI Desktop Stack

This deployment fork provisions bytebot-ui, bytebot-agent, Postgres, bytebot-desktop, Access Gateway. Application domains and any authored TCP endpoints are generated for each new deployment. Database and queue services are private by default.

Why Deploy Bytebot Protected AI Desktop Stack on Railway

New setup and maintenance documentation with attribution to the original template.
Public container sources are locked by image digest and public Git repositories by commit SHA.
Database and queue services use private networking; external database access is an explicit deployer choice.
A generated-password gateway protects the public UI and WebSockets while agent and desktop services stay private.

Common Use Cases

Automate a desktop task using your own model API.
Experiment with an AI agent controlling an isolated desktop.
Review the agent session through its web UI.

Dependencies for Bytebot Protected AI Desktop Stack

A Railway account with capacity for 5 service definitions and the authored storage.
Your own external accounts and credentials when an integration or model provider is needed. No paid provider subscription is included.
Local passwords and service secrets are generated or referenced at deployment; external integrations are configured by the deployer.

First Launch

Supply at least one supported model API key on bytebot-agent. Wait for the five services, then open Access Gateway. Log in as operator with PROXY_PASSWORD from its variables. The agent and desktop endpoints remain private.

Included Services and Storage

Service	Persistent paths
bytebot-ui	No attached volume
bytebot-agent	No attached volume
Postgres	`/var/lib/postgresql/data`
bytebot-desktop	No attached volume
Access Gateway	No attached volume

Operations

The gateway protects UI requests and WebSocket traffic. No model API credit is included. Do not store irreplaceable files in the desktop container; PostgreSQL persists agent records, but the desktop filesystem is ephemeral. Restrict who receives gateway access.

Railway bills the deployer for actual services, storage and network use. Back up persistent data before source/version changes. This publication was checked against Railway's configuration schema and upstream source/registry availability; a full runtime deployment test has not been performed.

Attribution and Source

Based on Bytebot by Bytebot. It appeared at position 16 in Railway's backend Popular/browse results captured on 2026-10-18, with 2,615 deployments reported for the original. That count belongs to the original, not this new fork.

The original software and repositories remain owned by their upstream authors and retain their licenses. This is a deployment configuration fork with new documentation, not a claim of software authorship or official vendor affiliation.

For template wiring issues, use this template's discussion link and share redacted logs. Never post passwords, API keys or login tokens.

Deploy this recipe

Review the required inputs and hosting resource costs before deploying.

Related deployment recipes

LibreChat Multi-Model Chat Starter — AI chat with MongoDB, private search and a vector-backed document API.
AI Provider Gateway Starter — 9Router with generated login, persistent config and your own AI providers.

These are separate deployments; no preconfigured integration between the applications is implied.

Disclosure: this template is maintained independently. Its maintainer may earn Railway usage kickbacks when other people deploy it. Railway bills hosting separately; no paid software subscription or provider credits are included.