> ## Content Index
> Fetch the complete content index at: https://nordisk-ai.ghost.io/llms.txt
> Use this file to discover other available public pages before exploring further.

# Anthropic Launches AI That Can Use Computers Like a Human
- URL: https://nordisk-ai.ghost.io/anthropic-launches-ai-that-can-use-computers-like-a-human/
- Published: 2024-10-27T15:46:39.000Z
- Updated: 2026-08-27T17:21:07.000Z
- Author: Anders Hagen
- Tags: AI Models, AI Security, AI Automation, Anthropic, Claude 3.5, computer use, office automation, #Migrated-1787841634512, #wp, #wp-post, #Import 2026-08-27 14:41, #Import 2026-09-07 13:41

Anthropic, known for its work in AI safety and research, has now unveiled advanced features in its latest AI model, Claude 3.5 Sonnet, along with the new model Claude 3.5 Haiku. The most groundbreaking news is the "Computer Use" feature, which lets Claude use a computer like a human - by navigating the screen, clicking buttons, and typing text. This development could dramatically change how companies automate complex workflows.

### An AI That Can Navigate Between Applications

The new "Computer Use" feature opens up many use cases in industries that depend on repetitive tasks. From data entry and research to customer service - the possibilities are broad and could have far-reaching effects on businesses. Mike Krieger, Chief Product Officer at Anthropic, explained in an interview with VentureBeat that the feature is still in beta, but that the technology could eventually improve data analysis and interactions between user interfaces, making many processes both faster and more efficient.

[Read More on Anthropic Here](null)

### Leading Companies See the Potential

Among Anthropic's first partners are GitLab, Canva, and Replit, all of which are already seeing the benefits of Claude 3.5 Sonnet. GitLab has started using the model to automate steps in its development processes and has seen a 10 percent improvement in reasoning ability, making the model well-suited to complex, multi-step workflows. Replit, a coding platform, sees potential in developing an AI-powered verification feature that could evaluate apps under development, potentially reducing bottlenecks in software development.

### What Does the "Computer Use" Feature Actually Do?

Claude differs from traditional automation tools in that it isn't tied to specific workflows. Instead, the model can "see" the screen through screenshots, interact with different applications, and adapt to new tasks. This flexibility makes the technology significantly more versatile than existing RPA solutions.

In a demo, Anthropic showed how Claude could fill out a form for Ant Equipment Co. without human involvement, by taking a screenshot, pulling the necessary information from a CRM database, and then filling out the form.

### Security and Privacy Concerns

Despite the technical advances, there are major security and privacy challenges. Anthropic has introduced several safeguards to reduce the risks. For example, Claude can only access a computer if a developer provides it with the necessary tools for screenshots and keystrokes. The new feature is also being released as a limited beta via API, so developers can test the technology in controlled environments.

### A New Era for Office Automation?

In the short term, companies could see productivity gains in areas like data entry, customer service, and IT support. But in the longer term, the technology could also come to be used in more complex fields, such as legal processes and medical diagnostics, where AI helps review contracts and analyze patient records.