TRENDING

Companies Have Wanted to Automate Repetitive Processes for Years. Anthropic’s AI Aims to Be the Solution They Need

The company’s new Computer Use function can autonomously recognize and interact with the PC interface to solve complex requests.

Anthropic
No comments Twitter Flipboard E-mail

Tightening screws is no fun, but for years, it was assembly line employees who handled the task. Eventually, robots replaced humans in many of these processes, and for a long time, companies have been seeking to do the same for computer-based tasks that are equally repetitive.

Robotic Process Automation. Robotic Process Automation (RPA) is a discipline that combines APIs with user interface interactions on computers. Its goal is to automate repetitive tasks, particularly in productivity and business applications.

Emulating human interaction. RPA enables the autonomous execution of various activities and transactions across different software systems. The idea is to replicate human user actions, allowing employees to focus on more complex tasks and thereby generate a greater return on investment.

Historically, RPA wasn’t AI. Until now, this type of automation hasn’t been considered artificial intelligence because RPA is process-based, while AI is data-centric. However, Anthropic’s new Computer Use feature has recently demonstrated that the two can be combined, allowing for the automation of user interface interactions to execute all kinds of repetitive tasks.

Anthropic 1 In the image, the Computer Use feature completes a form by taking information from a CRM system. Many companies’ dreams are coming true.

Anthropic’s perspective. In an interview with VentureBeat, Anthropic CEO Mike Krieger discussed the potential of the company’s new feature. He stated, “We anticipate it being particularly useful for tasks like conducting online research, performing repetitive processes like testing new software, and automating complex multi-step tasks.”

The question is: Who benefits most from automating complex tasks? That’s right–businesses.

A gift for businesses. The Computer Use feature can recognize what’s displayed on the screen and respond accordingly. This level of automation is particularly promising for sectors such as finance, legal services, and customer support. In this regard, Krieger added, “Claude could open spreadsheets, run analyses, and create visualizations. For customer service, it could navigate CRM systems to quickly find and update customer information.”

Massive potential. Features like these offer promising solutions for tasks such as data entry and customer support. However, automation can extend even further by recognizing and analyzing information displayed on the screen and acting upon it. This capability is what AI agents promise, and it serves as an exciting early example of their potential.

Limitations. Currently, Computer Use is available only on a limited basis, and Anthropic says it can’t use a PC as such. Instead, the tool operates within a controlled environment, enabling a range of tasks through autonomous, AI-generated mouse movements and keystrokes.

Nevertheless, Anthropic has implemented measures to detect if the feature is being misused for harmful activities, such as malicious interactions with government portals. While the tool has significant capabilities, it could also be exploited for nefarious purposes, similar to the risks posed by generative AI and deepfakes.

Image | Anthropic

Related | ‘Hello Computer’: Anthropic’s New AI Agent Offers a Glimpse Into the Future Envisioned by Star Trek in 1986

Home o Index