Amazon unveils Nova Act: AI model to control your browser and enhance Alexa
Amazon has revealed a new artificial intelligence model that enables users’ web browsers to be controlled by AI to perform tasks on their behalf.
Nova Act is a general-purpose AI agent that’s designed to take the digital assistant experience to the next level. The Nova Act SDK will enable developers to create agents that can execute complex workflows, simply by issuing a command to search for something and check out.
Right now, Nova Act is only available to developers looking to build apps that might make a restaurant reservation or order food. However, it’ll eventually come to Amazon’s Alexa voice assistant, which is the company’s most consumer-facing AI product.
While Nova Act is designed to enhance the Alexa experience, the Nova Act SDK is designed to help developers create agents that can work on behalf of end users. That’s similar to what China’s Manus AI has done with its own advanced AI agent capable of executing tasks independently, which has captured significant attention.
The Nova Act SDK will enable developers to create browser agents that can interact with web elements, as demonstrated by Nova Act’s ability to search for a location on Google Maps.
Amazon has also made its Nova foundation models available for public use at nova.amazon.com, which supports more than 200 languages and can handle contexts up to 300,000 tokens in length. The company plans to expand that limit to 2 million tokens. The company is also working on a reasoning model that it plans to release by the middle of 2025. Amazon is positioning itself as a competitor to OpenAI and Google.
While other companies have focused on creating agents for end users, Amazon’s strategy appears to be creating an entire AI stack, from foundational models through developer tools. The hope is that this will be a more efficient and scalable way of providing AI services in the long run.
That may not be as appealing to general consumers, but it may appeal more to organisations looking for deep integration of AI into their workflows. It also reflects a broader shift towards transparency and user empowerment in the development of AI.
