On Monday, Amazon unveiled Nova Act, a general-purpose artificial intelligence agent that can take control of a web browser and perform some simple actions on its own. Along with the new AI agent model, Amazon is releasing the Nova Act SDK, a toolkit that lets developers prototype agents using Nova Act.
On Monday, Amazon unveiled Nova Act, a general-purpose artificial intelligence agent that can take control of a web browser and perform some simple tasks on its own. Along with the new AI agent model, Amazon is releasing the Nova Act SDK, a toolkit that lets developers prototype agents using Nova Act.
Nova Act was developed by Amazon’s recently opened AGI Lab. It will also be the basis for key features in an upcoming update to Alexa+, a generative version of Amazon’s popular voice assistant enhanced with artificial intelligence. However, the version of Nova Act available today is a little less polished. Amazon calls it a research version, TechCrunch reports .
Developers can access Nova Act tools on the new website nova.amazon.com , which also serves as a showcase for Amazon's various Nova foundation models.
Amazon says developers using the Nova Act SDK should be able to automate basic actions on behalf of users, such as ordering salads at Sweetgreen or making reservations at a restaurant. With the Nova Act toolkit, developers can combine tools that allow an AI agent to navigate web pages, fill out forms, or select dates on a calendar.
Amazon claims that Nova Act outperforms OpenAI and Anthropic agents in several of the company’s internal tests. For example, on ScreenSpot Web Text, which measures how an AI agent interacts with text on a screen, Nova Act scored 94%, beating OpenAI’s CUA (which scored 88%) and Anthropic’s Claude 3.7 Sonnet (90%).
However, Amazon did not compare Nova Act using more common agent ratings such as WebVoyager.
Nova Act is the first publicly available product created by the aforementioned Amazon AGI Lab, an initiative co-led by former OpenAI researchers David Luan and Peter Abbil.
Luan says his team designed the Nova Act SDK to reliably automate short, simple tasks and give developers the tools to define exactly when they want a human to intervene in an agent's workflow. He hopes this will allow developers to build more robust agent applications, though not necessarily fully autonomous ones.
