Tag
Browser Use claims its V4 system is better than humans at browsing the web.
Cloudflare launches Kitesurf, a cloud-hosted browser built specifically for AI agents, enabling developers to automate web navigation and tasks without building browser software—available free in beta on its Browser Run platform.
WebRover is an MIT-licensed open-source AI agent that uses natural language to drive the browser for web automation, cross-site data scraping, and deep research, with support for local deployment.
Rindler is a product that automates web work teams currently do by hand, aiming to streamline repetitive online tasks.
Index is an open-source AI browser Agent developed by the Laminar team that turns any website into a callable API, supports vision models such as Claude and Gemini, and achieves 92% accuracy on WebVoyager.
Klura enables AI agents to turn website tasks into reusable capabilities, eliminating the need for repetitive UI crawling.
Nanobrowser is an open-source AI web automation tool, serving as a free alternative to OpenAI Operator. It runs locally in the browser, supports multi-agent collaboration, emphasizes privacy, and offers flexible LLM choices.
Apify and Coinbase launch x402 support, enabling AI agents to autonomously sign up, pay with USDC, and access over 20,000 web automation tools without human intervention.
The x402 protocol enables autonomous agents to pay for tools via USDC on Base, and Apify has expanded its x402 support from 2,000 to 20,000+ tools in partnership with Coinbase.
Apify and Coinbase have expanded x402 support to over 20,000 tools, enabling AI agents to autonomously pay per run using USDC on Base without requiring manual account setup or API keys.
An open-source agent that uses an LLM to control a real browser to fill forms and extract structured data, requiring minimal tokens per page.
BrowserAct is a web browser automation tool designed for AI agents, enabling automated web interactions.
A new paper introduces HLL, a benchmark for testing AI agents on real CAPTCHA tasks, showing that even strong agents fail on cluttered pages and recovery from mistakes.
Introduces SKILL.nb, a framework for governing reusable agent workflows through evidence-calibrated lifecycle policies, featuring selective formalization and gate-conditioned execution. It achieves significant improvements on web automation benchmarks and demonstrates resilience to environment drift.
This paper proposes SGDR (State-Grounded Dynamic Retrieval), an online skill learning method for web agents that enables stepwise, state-aware skill reuse rather than static task-level retrieval. Experiments on WebArena show SGDR achieves 37.5% success rate with GPT-4.1, a ~10.6% relative gain over strong baselines.
A comparison of Midscene and Browser-Use, two open-source tools with different focuses: Browser-Use is a web agent for one-time tasks, while Midscene is a vision SDK designed for reliable multi-platform repeated execution.
The author argues that AI agents fail on real web tasks not because models are weak, but because browsers are designed for humans and lack isolated, scriptable workspaces for agent use.
A fork of Playwright that generates unique browser fingerprints per session to enable AI agents to navigate the web undetected. The project is fully open-source under MIT license.
Browse.sh is a tool that gives AI agents muscle memory for automating web tasks.
Mini Browser is an agent-first browser CLI that enables AI agents to control a browser via Unix-style commands for navigation, scraping, screenshots, form filling, and more.