Amazon has unveiled Nova Act, a versatile AI agent that can control a web browser and perform some simple actions on its own. In the future, Nova Act will support all the features of Alexa+, Amazon’s updated voice assistant. Along with the agent, the company also released the Nova Act SDK, a toolkit that allows developers to create their own agent prototypes.
Image source: Pixabay
Nova Act is developed by Amazon’s newly opened San Francisco-based AGI Lab, led by former OpenAI researchers David Luan and Pieter Abbeel. Amazon calls the release of the AI agent a “research preview.” Developers can access the Nova Act toolkit now at nova.amazon.com, which also serves as a “showcase” for Amazon’s various Nova Foundation models.
The Nova Act is Amazon’s attempt to compete with the OpenAI Operator and Anthropic Computer Use with general-purpose AI agent technology. Many AI leaders believe that AI agents that can explore the web at the behest of users will make AI chatbots significantly more useful. Amazon is counting on the ubiquity of Alexa+ to give the new agent broad reach.
Developers using the Nova Act SDK will be able to automate basic actions on behalf of users, such as ordering groceries or booking a table at a restaurant. With Nova Act, developers can bundle tools that allow an AI agent to navigate web pages, fill out forms, or select dates on a calendar.
According to Amazon, Nova Act outperforms agents from OpenAI and Anthropic in several of the company’s internal tests. For example, in ScreenSpot Web Text, which measures how an AI agent interacts with text on a screen. Nova Act scored 94%, beating OpenAI’s CUA (88%) and Anthropic’s Claude 3.7 Sonnet (90%).
Experts say the main problem with recently released AI agents from OpenAI, Google, and Anthropic is their low reliability. They are slow in many tests, have difficulty making decisions on their own, and are prone to making mistakes that humans would not make. It will soon become clear whether Amazon has managed to rid its product of these shortcomings.
In March, Google Pixel 9 series smartphones received a new feature — real-time scam detection.…
Elon Musk's political activity after Donald Trump came to power in the United States was…
While the tactical strategy based on Star Wars from the American studio Bit Reactor and…
Google's Gemini Pro-powered AI notebook NotebookLM, currently only available for desktop users, is set to…
Experts have discovered a vulnerability in the WinRAR archiver that allows attackers to bypass the…
Analysis of Hubble telescope observation data for Uranus, the seventh planet in the Solar System,…