platoseed
The Open Source Agentic Browser
BrowserOS is an open-source agentic browser that runs AI agents locally.
BrowserOS is an open-source, privacy-first browser that runs AI agents locally to automate web tasks. It positions itself as a Chrome-like browser with built-in AI agents and extensive app integrations for local, end-to-end automation.
A Chromium-based open-source browser that ships with built-in AI agents running locally on the user's machine. Users describe tasks in natural language and the agent handles clicking, typing, and navigating, browsing the web, reading local files, and automating multi-step workflows within the browser. It supports multiple AI model providers (e.g., Claude, Gemini, OpenAI, Kimi) and can run local models via Ollama/LM Studio. It includes 40+ built-in MCP (message-crafted protocol) integrations to apps like Gmail, Slack, Notion, Calendar, etc., and can connect to custom MCP servers. Features include scheduled tasks, a pre-installed MCP server, and the ability to drive external tools via terminal through Claude Code or Gemini CLI. It emphasizes privacy with sandboxed agents and offline/local execution where possible, and is available on macOS, Windows, and Linux as a free/open-source product.
Who itβs for: Businesses and individual power users who want an open-source, privacy-conscious browser with built-in AI automation and a strong ecosystem of app integrations and local-first execution.
Backed by Y Combinator; 10k+ GitHub stars; active community with docs, FAQ, and dev tools; open-source contributions and Discord community.
Building open-source agentic browser. Privacy-first alternative to Perplexity Comet, Dia. Previously, ML Engineer at Google/Youtube and NVIDIA.
Building open-source agentic browser. Privacy-first alternative to Perplexity Comet, Dia.
Building open-source AI platform for next-generation AI hardware, reducing ML training costs by 30%.
Felafax AI launches a cloud layer and open-source ML stack to train across non-NVIDIA hardware (TPUs, Trainium, AMD, Intel GPUs) with out-of-the-box PyTorch XLA/JAX templates and a notebook-based setup for fine-tuning LLaMa 3.1 models; aims to deliver multi-TPU orchestration and cost-effective performance comparable to NVIDIA at 30% lower cost.
Formerly βFelafaxβ Β· why startups rename β

Leading open-source web agent project with 50k stars in 3 months

Operating system for the AI era