Anthropic, the AI startup backed by Amazon, founded by former OpenAI executives, announced Tuesday it has accomplished a new milestone in artificial intelligence: agents who can manipulate computers to get anything done in a way that strongly mimics human thinking.

A startup behind one of the popular chatbots that competes against OpenAI's ChatGPT and Google's Gemini is rising to fame. Others include Anthropic, which competes in an extremely fast and intensifying race with Google, Amazon, Microsoft, and Meta, in the generative AI market, which is expected to generate over $1 trillion in revenue by the end of the next decade.

The first of the two new AI models revealed bears the name "Computer Use," or, as Anthropic's branded it, a breakthrough. This tech gives its AI the capability to interpret what is on a computer screen, pick buttons, enter text, and browse the internet, to name a few things, across different platforms.

According to CNBC, Jared Kaplan, chief science officer at Anthropic, described the way the tool works: "It can use computers in basically the same way that we do," while pointing out that the system is capable of executing "tens or even hundreds of steps" in order to perform the task.

The tool has been available to early access partners, including Amazon and companies like Asana, Canva and Notion. Anthropic has actually been working on it since the beginning of the year and rolled it out to developers in a public beta on Tuesday. Kaplan hopes that the company extends access to consumers and businesses over the coming months or by early next year.

Its potential consumer uses might include booking flights and appointments, filling out forms, doing a bit of online research, and filing expense reports. "We want Claude to help people with all kinds of work," Kaplan said, noting that while chatbots answer questions, "the current setup is limited" in the sense that it doesn't provide much more context or action than that.

AI agents like those developed by Anthropic can be much more ambitious than chatbots because they can provide answers to much more complex, multi-step queries. Such agents are not just response-providing tools but will execute tasks on behalf of users, potentially transforming the way business is done. As Kaplan pointed out, these agents will emerge as virtual assistants to "virtual collaborators."

There has been significant momentum in the push toward AI agents across the tech world. Microsoft CEO Satya Nadella reportedly said that his company is also working toward creating AI agents to help users do more things and will be able to overcome the challenges. Executives from Meta and Google have similar ambitions regarding making AI agents more productive.

Since it was founded, Anthropic has rapidly become a significant force in the AI landscape. It brought to market its Claude chatbot last March, right in OpenAI's court, as a business-to-business and business-to-consumer product. Among its backers are the likes of Google, Salesforce, and Amazon, though yet not allowed to much of the consumer market.

Anthropic recently launched iOS and Android apps, plus a "Team" plan for businesses, opening it further to European markets. Its latest product is Claude Enterprise, for companies who wish to integrate AI into their functioning. Some of its early adopters include GitLab, Midjourney, and Menlo Ventures.

The Claude Enterprise product offers customers a wider contextual window to upload data so that businesses can process the equivalent of 100 sales calls lasting 30 minutes, 100,000 lines of code, or 15 full financial reports at a time. The product also has "activity feeds," which enable AI veterans within companies to show new users how the technology is being used.

Nor was that the end of the road for Anthropic in terms of progress toward AI. June marked the launch of "Artifacts," or "workspaces," where users can create text documents and code and then edit these real time with Claude's guidance. The system enables the enterprise client to build marketing plans, process sales data, draft legal documents, and much more.

The company also hired other industry influencers. For example, in May, Mike Krieger, co-founder and former chief technology officer of Instagram, came on as Chief Product Officer. He led Instagram to 1 billion users, meanwhile the engineering team here grew over 450. Jan Leike, formerly a safety lead at OpenAI, joined Anthropic in May.

The latest announcements from Anthropic carry the flavor of longer-term ambitions to be molded over time into building AI agents that function as more than assistants-only but, in turn, become highly capable virtual collaborators in business and beyond.