Unify customer data, automate workflows, and scale experiences with Einstein AI, Copilot, and Data Cloud in one CRM platform.
110M visits/mo+1.1%#1 Business & Marketing
Nabsun
Open-source AI browser that assists with browsing tasks using local or cloud models
About Nabsun
Nabsun is an open-source browser with a built-in AI assistant designed to handle routine browsing tasks such as reading pages, extracting key ideas, answering questions, and performing actions like clicking, typing, and navigating within existing tabs. It operates on either local models running on the user’s CPU or cloud-based models via API keys, allowing flexibility in model choice and cost. The assistant defaults to requesting approval before taking actions but can be configured for Autopilot mode for routine tasks. Nabsun integrates with existing workflows by functioning as an MCP server, enabling tools like Claude Code or Codex to interact with the browser through the same login and approval settings. The browser retains familiar features such as encrypted passwords, bookmarks, downloads, and extensions while adding AI-driven assistance. A local configuration file, soul.md, stores user preferences like name and time zone to personalize interactions without exposing data to cloud providers by default. The built-in model is optimized for everyday tasks, while larger models can be used for complex research.
Key features
- Built-in AI assistant for reading and summarizing pages
- Local model included (Qwen3 1.7B parameters)
- Approval-based action system with Autopilot mode
- MCP server for editor integration (Claude Code, Codex)
- Supports cloud models via API keys (Anthropic, OpenAI)
- Local model integration with Ollama
- Personalization via soul.md configuration file
- Familiar browser features (passwords, bookmarks, extensions)
Use cases
- Automating routine web research across multiple tabs
- Filling forms and navigating websites with AI assistance
- Integrating browser actions into editor workflows via MCP
Pros
- Open-source and locally runnable
- Supports both local and cloud-based AI models
- Integrates with existing browser tabs and logins
- Includes built-in local model with no account required
- Functions as an MCP server for editor integration
Cons
- Unsigned Windows build requiring manual override
- Developer preview with potential rough edges
- Local model limited to CPU-only operation
- No official macOS or Linux builds mentioned
Nabsun videos
Frequently asked questions about Nabsun
What is Nabsun?
Nabsun is an open-source AI-powered browser with a built-in assistant that automates routine browsing tasks such as reading pages, extracting key ideas, answering questions, and performing actions like clicking, typing, and navigating within existing tabs.
Who is Nabsun designed for?
Nabsun is designed for users who want to reduce busywork in their browser workflows, such as researchers, developers, or anyone who frequently interacts with web-based tools and needs assistance with repetitive tasks.
How does Nabsun handle model selection and costs?
Nabsun allows users to run models locally on their CPU at no cost or connect to cloud-based models via API keys, providing flexibility in model choice and cost management without model lock-in.
Can Nabsun integrate with other tools like editors or IDEs?
Yes, Nabsun functions as an MCP server, enabling tools like Claude Code or Codex to interact with the browser through the same login and approval settings, bridging your editor and browser workflows.
Does Nabsun require an internet connection to work?
The built-in local model in Nabsun can operate entirely offline, while cloud-based models require an internet connection for API access.
How does Nabsun ensure user control and privacy?
Nabsun defaults to requesting approval before taking actions, and user preferences are stored locally in a file called soul.md, which is not shared with cloud providers by default, ensuring privacy.
