In a groundbreaking development, Anthropic's Claude 3.5 Sonnet large language model (LLM) has gained the ability to operate computers, opening up new opportunities in robotic process automation (RPA) and beyond. This "computer use" capability, currently in beta test, enables developers to instruct Claude 3.5 Sonnet to read and interpret on-screen information, type text, move the cursor, click buttons, and switch between windows or applications.

This breakthrough has significant implications for the RPA market, as Claude 3.5 Sonnet's ability to understand and interact with computer environments like a human user could shake up the industry. UIpath, a leading RPA vendor, has already integrated Claude 3.5 Sonnet into three of its products, demonstrating the potential for widespread adoption.

While the technology is still in beta and comes with limitations, such as struggles with high-resolution screens and risks of prompt-injection attacks, the possibilities are vast. As Paul Chada, co-founder of AI startup Doozer AI, notes, Claude's computer use capability addresses core challenges in RPA, including adaptive interaction and working across any interface.