Using AgentKit with Browserbase

Develop AI Agents that can browse the web autonomously using Browserbase and AgentKit (by inngest).

Overview

By the end of this guide, you’ll have an AI agent built with AgentKit that can browse the web using Browserbase.

You’ll learn how to:

1. Create AgentKit tools that leverage Browserbase’s managed headless browsers
2. Build autonomous web browsing agents that can search, extract data, and interact with websites
3. Use Stagehand, Browserbase’s AI-powered browser automation library, to create resilient web agents

This integration is useful for:

• Building AI agents that can fetch real-time information from the web
• Creating autonomous browsing capabilities within your agents
• Developing resilient web scraping and interaction systems
• Enabling your agents to perform complex web-based tasks

There are two approaches to using Browserbase with AgentKit:

1. Create custom Browserbase tools: Ideal for simple actions on webpages with manual browser control.
2. Use Stagehand library as tools: Better for autonomous browsing and resilient web scraping.

Prerequisites

Before you start, make sure you have:

• AgentKit installed
• Browserbase Project ID & API key
• (Optional) LLM API key of your choice to use with Stagehand

Next, let’s dive into building web browsing agents with AgentKit and Browserbase.

Quickstart Guides

Autonomous Browsing with Stagehand

Leverage Stagehand to build fully autonomous web browsing agents that can navigate, observe, extract data, and interact with websites.

Reddit Search Example

Explore a complete example of a Reddit search agent built with AgentKit and Browserbase.