Anthropic AI startup supported by Amazon and cofounded by former OpenAI executives announcing an important milestone in AI. It has recently released AI agents that can do extremely complex tasks on a computer accurately as a human would through its new “Computer Use” capability. This feature allows AI to interpret what is going on on a computer screen, click buttons, enter text, navigate websites, and execute tasks across all software platforms in real time.
This technology is built into two of Anthropic’s newest AI models and was announced in conjunction with the public beta of its availability for developers. Companies such as Amazon, Asana, Canva, and Notion were among the early testers of the tool, having been granted access to it before its broader rollout. Anthropic’s chief science officer, Jared Kaplan, said that, in one demonstration of the tool, it could complete tasks that involve “tens or even hundreds of steps.” This makes it an incredibly powerful tool in terms of what it is doing for digital productivity.
Part of a broader trend in the artificial intelligence sector, AI agent development is in line with increased urgency by startups and tech giants like Google, Microsoft, Meta, and Amazon in their mad dash to design the most developed generative AI systems. This is far from simple chatbots or content generators. They are programmed to finish tasks on behalf of users with such intricate interdependence that it gives them more productivity.
AI agents, unlike those in other forms of chatbots, can be designed and configured towards specific business functions. For instance, this kind of AI assistant personifies the fictional AI that Tony Stark has for himself in the Marvel Universe-known as J.A.R.V.I.S.
Released to market in March 2023, Anthropic’s AI assistant continues to gain momentum in the marketplace directly competing with both OpenAI’s ChatGPT and Google’s Gemini. Competitive and now boasting Google, Salesforce, and Amazon as major investors, Anthropic continues its growth through additional product releases. During this year alone, the company has introduced iOS and Android apps, a team plan for businesses, and launched in Europe.
The biggest news about Anthropic was the “Claude Enterprise,” a business unit particularly designed for businesses. Claude Enterprise has a much broader context window. Thus, clients can upload and then process vast amounts of data, for example, 100,000 lines of code or 15 financial reports. It is very handy for companies that carry out complex tasks: drafting legal documents, writing codes, or analyzing sales data. Another feature that is being offered is “Artifacts,” a way to create text or code in partnership with others, work on it together in real-time, develop marketing materials, automate tasks, and summarize documents.
That perhaps, underpinning the Anthropic vision lies in the manner in which AI agents can be made to become more virtual collaborators rather than assistants. The company wants to develop AI that would help people book flights, conduct online research, or even manage appointments. Kaplan foresees the potential wherein Claude becomes a utility that helps them with work-related problems in every single way imaginable, unlike current uses of typical chatbot interfaces.
The competition in developing more complex AI agents on an increasing scale has upped the ante by increasing the involvement of Anthropic with their latest innovation putting it in a tight spot in competition within the rapidly growing generative AI marketplace.
