Run and manage your own AI agents on local or remote servers
Imagine accessing an AI agent from a browser, desktop, or mobile device that executes code and handles files. Rakazo is an open-source tool that lets you host and manage such agents on your own server or local machine. It is not tied to a specific cloud service, allowing you to choose and combine your preferred models and execution environments.
Persistent memory and isolated workspaces
The core of Rakazo is the 'persistent bot'. Each bot maintains its own conversation history, memory, routines, and history, retaining context over time. Bots can access the browser, terminal, file system, and graphical desktop, enabling them to perform actual tasks beyond simple chat. They can also delegate work to other bots or create short-lived sub-agents to distribute complex tasks. Execution environments can be selected from Docker, E2B, Daytona, CreateOS, Boat, or a trusted local computer, allowing you to adjust security and performance as needed.
Custom models and voice integration
Users can directly connect their desired LLM model credentials via Pi. This means you are not dependent on a specific vendor and can freely switch between various AI models. Voice mode is also supported, allowing you to listen to bot responses, give voice commands, or make calls by choosing from ElevenLabs, OpenAI, Cartesia, or Fish Audio. App integrations are possible through Composio or Pipedream Connect, and it also supports Treg, remote MCP, and OpenAPI tool sources installed by the user. This flexibility allows Rakazo to function as extensible AI infrastructure rather than just a chatbot.
Installation and environment setup
The fastest start is provided by running the supplied installation script in an environment with Docker Engine, the Compose plugin, curl, and OpenSSL installed. This process downloads the Compose file, generates a random secret in .env, and starts Rakazo without cloning or installing Node.js. The local Docker computer is enabled by default, and remote providers such as E2B, Daytona, CreateOS, Boat, or Railway can be connected with API keys if needed. To run on a server, use the same installer on a VPS and enter the server address in the desktop or mobile app to connect. HTTPS configuration must be performed via Caddy as instructed in the documentation.
Considerations for the beta stage
Rakazo is currently in beta and is licensed under the Apache License 2.0. For local development, checking out the source requires Node.js version 22.22.2 or higher (22.x series), 24.x, or 26+, along with pnpm 9 and Docker. Node.js 23.x and 25.x are not supported, so you must verify the version. The database uses PostgreSQL, with migrations managed through Prisma. Security secrets must be set in .env using a value generated by openssl rand -hex 16, and Docker sandboxes require a dedicated SANDBOX_SUPERVISOR_TOKEN. If using Treg, usage is billed, and embedding in hosted products may require a written contract, so terms should be reviewed.