Anthropic, an Amazon-backed company, introduces AI agents capable of performing intricate tasks, competing with OpenAI, Microsoft, and Google.

• The Amazon-backed AI startup, Anthropic, founded by former OpenAI research executives, unveiled AI agents capable of performing intricate tasks like humans.
• AI agents are designed to optimize productivity and handle intricate, multi-step tasks on behalf of users.
• AI models can be customized for specific business functions.

The Amazon-backed AI startup, Anthropic, announced on Tuesday that its AI agents have achieved a milestone, enabling them to complete complex tasks like humans do using a computer.

Anthropic, the company behind Claude, one of the chatbots that has gained immense popularity like OpenAI's ChatGPT and Google's Gemini, is part of a generative AI arms race with startups and tech giants such as Meta, Microsoft, and Google to ensure they remain competitive in a market predicted to reach $1 trillion in revenue within a decade.

Anthropic's latest AI models include a new Computer Use capability that enables its technology to interpret screen content, click buttons, enter text, navigate websites, and execute tasks through any software and real-time internet browsing.

Jared Kaplan, Anthropic's chief science officer, stated in an interview with CNBC that the tool can perform tasks with "tens or even hundreds of steps" using computers in a way that is similar to how we use them.

According to CNBC, Amazon had early access to Anthropic's tool and early customers and beta-testers included Asana, Canva, and Notion. The company has been working on the tool since early this year, as stated by Kaplan.

Anthropic made the feature available in public beta for developers on Tuesday. The team plans to make it accessible to consumers and enterprise clients in the near future, either in the next few months or early next year, as per Kaplan.

Anthropic stated that future consumer applications will involve booking flights, scheduling appointments, completing forms, conducting online research, and submitting expense reports.

Kaplan stated to CNBC that the chatbot setup is limited in its ability to assist people with various types of work as it only provides context after answering a question.

What is an AI agent?

The industry has shifted from text-based AI responses to AI-generated photos, videos, and voice after the success of OpenAI's ChatGPT. Now, startups and Big Tech companies are fully investing in AI agents.

AI agents are designed for productivity and can complete complex tasks on behalf of users, unlike chatbots. They are typically customized for specific business functions and can be built on large AI models. J.A.R.V.I.S., Tony Stark's AI assistant from the Marvel Universe, is an example of an AI agent.

In June, Grace Isford, a partner at venture firm Lux Capital, stated that there has been a significant increase in interest among tech investors in startups developing AI agents. These companies have collectively raised hundreds of millions of dollars and experienced a rise in their valuations, in line with the broader generative AI market.

Microsoft CEO Satya Nadella stated on an earnings call that he aims to develop an AI agent capable of performing more tasks for users, although there is still much work to be done in execution. Similarly, executives from Meta and Google have highlighted their progress in making AI agents more productive.

Anthropic is competing with OpenAI on multiple fronts

Since its release of Claude in March 2023, Anthropic has become a highly sought-after AI startup, competing directly with OpenAI's ChatGPT in both the enterprise and consumer markets. The company has gained significant backing from major investors such as Google, Salesforce, and Amazon. In January, Anthropic introduced iOS and Android apps, a Team plan for businesses, and expanded internationally into Europe.

In September, Scott White, a product manager at Anthropic, told CNBC that the models will increasingly behave like virtual collaborators rather than virtual assistants in the future.

Anthropic's Tuesday announcements are a crucial part of its plan to develop virtual collaborators, or agents.

Anthropic recently launched Claude Enterprise, its largest new product since its chatbot's introduction, aimed at businesses seeking to incorporate Anthropic's AI. The beta testers and early clients for the enterprise product were GitLab, Midjourney, and Menlo Ventures, as stated by the company.

Claude Enterprise offers a larger context window for uploading relevant documents, equivalent to 100 30-minute sales conversations, 100,000 lines of code, or 15 full financial reports, according to Anthropic. Additionally, the plan includes activity feeds for super-users to demonstrate AI usage to newcomers within a company, as stated by White.

Anthropic's June debut of its more powerful Claude 3.5 Sonnet and May rollout of its "Team" plan for smaller businesses were followed by the Claude Enterprise launch.

Anthropic unveiled "Artifacts" in June, a feature that lets users request specific tasks from its Claude chatbot, such as generating a text document or code, and then opens the output in a separate window.

White stated in September that Anthropic's enterprise-level clients can utilize "artifacts," or "workspaces," to view, edit, and build upon Claude's creations in real-time. These workspaces will enable clients to create marketing calendars, input sales data, create dashboards or forecasts, draft code for features, write legal documents, summarize complex contracts, and automate legal tasks.

In May, Anthropic debuted Teams and welcomed Mike Krieger, co-founder and former CTO of Meta-owned Instagram, as its chief product officer. Under Krieger, the platform grew to 1 billion users and its engineering team expanded to over 450 members. Additionally, OpenAI's former safety leader, Jan Leike, joined the company that same month.