Amazon’s Nova Act is a new AI tool that can control web browsers without human help. Part of the upcoming Alexa+ upgrade, it performs online tasks like ordering food and making reservations. The technology processes text, images, and videos to understand diverse information. It comes in three versions: Micro, Lite, and Pro, with varying processing capabilities. Built-in safety features protect users from misinformation. Further details reveal impressive technical specifications.
Amazon has revealed its latest AI technology called Nova Act, which can control web browsers by itself. This new AI agent is part of Amazon’s upcoming Alexa+ upgrade that will include more advanced AI features. The company’s AGI lab in San Francisco developed this technology, which they claim performs better than similar tools from competitors like OpenAI and Anthropic.
Nova Act can handle simple online tasks without human help. It can order food, make restaurant reservations, and complete other basic web actions. The AI works with text, images, and videos, allowing it to understand different types of information. Developers can access the Nova Act SDK at nova.amazon.com to build their own applications. The project is being led by former OpenAI researchers David Luan and Pieter Abbeel who focus on creating AI systems that mimic human computer use. Nova Act incorporates data product approach to ensure high-quality outputs by treating data as reusable, governed assets. The technology utilizes natural language processing capabilities that allow it to interpret human speech and text effectively.
Amazon’s Nova Act automates everyday web tasks like food ordering, handling multiple data formats while offering developer access via SDK.
The technology comes in three main versions. Nova Micro can process up to 128,000 tokens of information at once. The more powerful Nova Lite and Pro models can handle 300,000 tokens. All models support over 200 languages and can generate responses quickly, with Nova Micro creating more than 200 tokens per second.
Amazon built safety features into Nova Act to protect users. The AI includes content moderation tools and safeguards against misinformation. It also allows for human oversight when needed. Amazon provides transparency through AWS AI Service Cards that explain how the system works and its limitations.
In the competitive AI market, Amazon claims Nova Micro beats Google’s Gemini 1.5 Flash and Meta’s Llama 3.1 8B in certain tasks. The company is trying to solve reliability problems that affected early AI agents from other companies.
Looking ahead, Amazon plans to release two more Nova models in 2025. These will include a speech-to-speech model for natural conversations and an “any-to-any” model that can work with many types of content. The company says it’s working with academic researchers to continue improving its AI technology while keeping it responsible and safe for users.
