It was the spring of last year that Anthropic announced the development of artificial intelligence in order to strengthen virtual assistants; Virtual assistants who can conduct research, respond to emails, and perform support work on their own.
The company called the algorithm “the next-generation algorithm for self-learning AI,” one that it believed could one day automate a large part of the economy if all goes according to plan.
Anthropic is an American artificial intelligence startup founded by former members of OpenAI. Entropic is engaged in the development of general artificial intelligence systems and language models with responsible use of artificial intelligence.
It’s been a long time since Enthropic news and it looks like their AI is launching; The company on Tuesday released an upgraded version of its Claude 3.5 Sonnet model that can understand and interact with any desktop application.
Through a new API, the model can mimic keystrokes, button clicks, and mouse movements, essentially simulating a person sitting behind a PC.
“We trained the cloud to see what’s happening on the screen and then use available software tools to perform tasks,” Entropic said of the development. When a developer commissions and grants access to the cloud using a desktop application, the cloud looks at screenshots of what is visible to the user, then counts how many pixels it takes to move and click. It needs vertical or horizontal.
Developers can try computing through Anthropic API, Amazon Bedrock, and the Google Cloud Vertex AI platform.
A tool that can automate tasks on a PC is not a new idea; Countless companies offer such tools.
In the race to develop so-called “artificial intelligence agents,” the field has become more crowded; AI agents generally refer to artificial intelligence that can automate the operation of software.
Some analysts say AI agents could provide an easier path for companies to monetize the billions of dollars pouring into AI.
According to a recent Capgemin survey, ten percent of organizations are currently using AI agents, and 82 percent plan to do so within the next three years.
Among the companies that have taken the plunge is Salesforce, which made some big announcements about its AI agent technology this summer.
Microsoft also announced new tools for building artificial intelligence agents yesterday. OpenAI is also designing its own brand of artificial intelligence agents.
Startup Rabbit is building a web agent that can do things like buy movie tickets online. Adept, recently acquired by Amazon, trains models to browse websites and navigate software, and Twin Labs uses models including OpenAI’s GPT-4o to automate desktop processes.
Anthropic claims that Sonnet 3.5 is a powerful model that performs coding tasks even better than OpenAI’s flagship O1, according to the SWE-bench Verified benchmark.
Despite not being explicitly trained to do so, the enhanced Sonnet 3.5 will correct and retry its tasks if it encounters obstacles.
Entropic calls its take on the AI agent concept an “execution layer” that allows the new Sonnet 3.5 to execute desktop-level commands. 3.5 Sonnet can use any website and any application thanks to its web browsing capabilities.
In an evaluation designed to test the ability of an AI agent to assist with airline reservation tasks, the new Sonnet 3.5 managed to complete less than half of the tasks successfully.
Latest Passing over countries : Spain | Dominica | United Arab Emirates
RCO NEWS
