Show HN: BrowserOS -- browser agents with GPT-OSS, local llms

BrowserOS is an open-source Chromium fork enabling non-developers to create and run browser agents locally. After testing three UX approaches (drag-and-drop workflows, one-shot agents, and plan-follower agents), the team found that having users provide simple natural language plans while the LLM executes each step achieved the best balance—improving success rates from 30% to ~80% even with local models like those run via Ollama or LMStudio.

Key Points

• •BrowserOS is an open-source Chromium fork enabling non-developers to create and run browser agents locally with privacy-first support for local LLMs
• •One-shot agents (high-level task only) achieved only ~30% success rate due to poor planning by smaller local models and unreliable execution
• •Drag-and-drop workflow approach (like n8n) was too complex and time-consuming (20+ minutes setup), making it worse UX than manual task completion
• •Plan-follower agents emerged as optimal middle ground: users provide simple natural language steps, LLM executes each step without planning responsibility
• •Plan-follower approach achieved ~80% success rate with local models by reducing LLM cognitive load to execution-only tasks
• •Agent builder auto-generates initial plans that users edit/customize, reducing friction from 0 to 30 seconds of user effort
• •Privacy-first architecture supports local LLMs via Ollama and LMStudio, eliminating cloud dependency and data exposure
• •Key UX insight: reliability and ease-of-use require balancing human guidance with AI execution rather than full automation or manual configuration

Workflow Diagram

Concepts