GitHub

Project Description: surf.new

What is the project about?

surf.new is a playground for testing and visualizing web agents powered by Steel.dev. It allows users to see how different AI agents interact with websites in real-time, simulating human-like browsing behavior. It's essentially a testing and demonstration platform for AI-powered web automation.

What problem does it solve?

The project addresses the challenge of testing and understanding how AI agents interact with the web. It provides a controlled environment to:

  • Visualize agent behavior: See exactly how an agent navigates a website, clicks elements, and fills out forms.
  • Compare different agents: Test various AI providers (like Claude and GPT-4) side-by-side to see their strengths and weaknesses in web interaction.
  • Debug agent interactions: Identify issues and refine agent logic by observing their actions in real-time.
  • Demonstrate capabilities: Showcase the potential of web agents for various tasks.

What are the features of the project?

  • Multi-Agent Testing: Supports testing multiple web agents simultaneously.
  • Real-time Visualization: Provides a live view of the browser as the agent interacts with it.
  • AI Provider Support: Integrates with various AI providers, including Claude and GPT-4.
  • Modern Web Technologies: Built using a modern stack for performance and maintainability.
  • User-Friendly Interface: Offers a clean and intuitive UI for easy interaction.

What are the technologies used in the project?

  • Frontend:
    • Next.js 15 (React Framework)
    • Shadcn UI (UI Components)
    • Tailwind CSS (Styling)
    • Vercel AI SDK (AI Chat Interface)
  • Backend:
    • FastAPI (Python Web Framework)
    • Steel SDK (Browser Automation)
    • Langchain (Agent Framework)
  • Other:
    • Python 3.11+
    • Node.js 18+
    • npm

What are the benefits of the project?

  • Improved Agent Development: Facilitates the development and refinement of web agents.
  • Enhanced Understanding: Provides insights into how AI agents interact with the web.
  • Easy Testing and Comparison: Simplifies the process of testing and comparing different agent implementations.
  • Open Source and Extensible: Allows for community contributions and customization.
  • Demonstrates Steel.dev Capabilities: Showcases the power of the Steel.dev platform.

What are the use cases of the project?

  • Web Automation Testing: Testing and debugging web automation scripts and agents.
  • AI Research: Studying and comparing the behavior of different AI models in web environments.
  • Educational Tool: Learning about web agents and how they work.
  • Proof-of-Concept Development: Demonstrating the feasibility of web-based AI applications.
  • Competitive Analysis: Benchmarking different AI agent providers.
  • Task Automation Development: Prototyping and refining automated workflows for web-based tasks.
surf.new screenshot