bytebot

Bytebot is a desktop AI agent that ships with its own virtual desktop and an integrated task management stack. It exposes a REST and UI-driven interface to create and monitor tasks, and provides programmatic controls to interact with the desktop (for example, taking screenshots or performing automated UI actions). The system is designed to operate with multiple AI backends (Anthropic, OpenAI, Gemini) and can authenticate to websites and services via installed password managers. To use Bytebot, deploy the Docker setup described in the Quick Start, configure your AI provider keys in the docker/.env file, and then access the UI and APIs exposed by Bytebot. You can create tasks in natural language, upload documents for processing, and watch Bytebot perform multi-step workflows across desktop applications and web portals. For automation, you can submit tasks via the API (POST /tasks) and issue desktop actions (e.g., screenshot or click_mouse) using the computer-use endpoints to drive interactions on the virtual desktop.

Prerequisites:

• Docker and Docker Compose installed on your machine
• Git installed
• Optional: API keys for Anthropic, OpenAI, or Google Gemini if you plan to enable external AI backends

Installation steps:

1. Clone the repository: git clone https://github.com/bytebot-ai/bytebot.git cd bytebot

2. Configure AI provider keys:

   • Create a file docker/.env and add one or more of the following keys, depending on your provider: ANTHROPIC_API_KEY=sk-... OPENAI_API_KEY=sk-... GEMINI_API_KEY=...
   • Example: echo "OPENAI_API_KEY=sk-..." > docker/.env

3. Start Bytebot with Docker Compose: docker-compose -f docker/docker-compose.yml up -d

4. Open the Bytebot UI (desktop view) in your browser: http://localhost:9992

Notes:

• You can stop Bytebot with: docker-compose -f docker/docker-compose.yml down
• The docker-compose.yml in the docker directory wires up the virtual desktop, agent, and UI components and expects the environment keys from docker/.env.

Tips and common issues:

• Ensure Docker has enough resources allocated (CPU, RAM) for a smooth virtual desktop experience.
• If you change provider keys, restart the containers to pick up new credentials.
• The API endpoints include:
  • POST /tasks to create tasks (with optional file uploads)
  • POST /computer-use to perform desktop actions like taking a screenshot or simulating mouse clicks
• If you encounter port conflicts, verify that ports 9992 (UI) and 9990/9991 (API) are free or adjust docker-compose.yml accordingly.
• For persistent environments, Bytebot installs programs inside its virtual desktop so they remain available across tasks.
• Refer to the Full deployment guide for alternative deployment options (Railway, etc.).