HappyCapy is a browser-based workspace that gives you an agent running on its own cloud computer. Each agent works in an isolated sandbox where it can browse the web, handle files, and run code, and you can watch what it does and stop a task at any point. Tasks you hand off keep running when your device is offline, and finished work lands in an inbox.
Capabilities come from skills: packaged integrations and workflows from a Skill Store that cover research, image and video generation, slide building, web development, and connections to your apps. You can also create your own skills and set up recurring automations, such as a daily research dashboard or a weekly email digest.
A single subscription covers access to more than 20 frontier models from several labs, including text models from the Claude, GPT, DeepSeek, Grok, Kimi, and Qwen families and media models such as Seedream, Seedance, Wan, and Nano Banana. You switch models per task without managing API keys.
Features
- Cloud agent sandbox: isolated environment per agent with browsing, file access, and code execution
- Async task inbox: hand off work that continues while you are offline and collects results in one place
- Live observation: see the agent's actions as they happen and interrupt any task
- Skill Store: installable skills for research, media generation, slides, site building, and app connections
- Custom skills and automations: author your own skills and schedule recurring jobs
- Model switching: 20+ text and media models under one subscription, no API keys required
- Media generation: images, videos, and slide decks produced by the same agent
