Yazılar

Amazon Unveils Nova Act AI Agent in Research Preview, Expands Access to Frontier AI Models

/in /tarafından

Amazon has unveiled a new artificial intelligence (AI) agent called the Nova Act, designed to perform various browser-based tasks. Powered by the company’s Nova family of AI models, this new AI agent is a significant leap forward in Amazon’s efforts to provide “frontier intelligence” with competitive pricing. The Nova Act is currently available in a research preview and is accessible to users in the United States. It comes as a software development kit (SDK), allowing developers to explore and integrate its agentic capabilities into their applications and systems.

The Nova family of AI models was first introduced in December 2024, with three variants—Micro, Lite, and Pro—tailored to different performance needs. These large language models (LLMs) are designed to offer advanced intelligence, with Amazon emphasizing their affordability and superior performance. Initially, access to these models was restricted to invite-only businesses, but with the launch of the Nova Act, Amazon is expanding availability to developers and AI enthusiasts. Users can now explore the Nova models on a dedicated website, which also features the Nova Act AI agent.

The Nova Act AI agent is designed to automate a variety of tasks within a web browser. While Amazon has not fully detailed the range of tasks the AI agent can perform, it is expected to handle tasks such as filling out online forms, shopping, or even booking restaurant reservations. This makes it a versatile tool for users looking to streamline their digital tasks. The AI agent is built on a custom version of the Nova models, which allows it to perform these tasks autonomously, offering a glimpse into the future of agentic AI systems capable of acting on behalf of the user.

Amazon has positioned the Nova Act as part of its broader vision for AI agents—systems that can handle tasks across both digital and physical environments. By releasing it as an SDK and research preview, the company is encouraging developers to experiment with this new technology and explore its potential applications. With this move, Amazon is not only expanding the reach of its AI models but also pushing the boundaries of what AI can achieve in terms of real-world, browser-based task automation.

Scale AI Eyes $25 Billion Valuation in Potential Tender Offer Amid AI Boom

/in /tarafından

Scale AI, a rapidly growing artificial intelligence startup, is seeking a valuation as high as $25 billion in a potential tender offer, according to a report from Business Insider. The company, which has seen rising demand for AI services, aims to leverage the current AI boom, which has attracted significant attention from major tech players.

The startup, based in California, is in discussions regarding the terms of the tender offer, which could involve investors or the company purchasing shares from existing shareholders. However, the final valuation may change depending on negotiations.

Founded in 2016, Scale AI is already valued at nearly $14 billion, as per its last funding round. The company counts tech giants such as Nvidia, Amazon, and Meta among its backers. Scale AI specializes in providing high-quality, accurately labeled data essential for training machine learning models, including advanced tools like OpenAI’s ChatGPT.

Despite its success, the company is currently under investigation by the U.S. Department of Labor regarding compliance with the Fair Labor Standards Act, adding a layer of regulatory scrutiny.

Amazon Set to Move Alexa Voice Processing to the Cloud, Discontinuing Local Processing

/in /tarafından

Amazon is reportedly notifying Echo device users that, starting March 28, the option for local processing of voice requests will be discontinued. The move comes as the company shifts its focus to its new AI-powered version of Alexa, known as Alexa+. Unlike the previous iteration, which allowed on-device processing of voice recordings, Alexa+ will rely entirely on cloud-based processing to handle voice interactions. Users who opt to keep their devices configured for local processing will lose access to certain features, including Voice ID functionality, which helps Alexa recognize individual voices.

This decision marks a significant change from Amazon’s approach in 2021 when it introduced on-device processing for voice requests as a privacy-focused option. At the time, the feature was meant to give users more control over their conversations with Alexa by allowing voice commands to be processed locally, without sending data to the cloud. However, according to recent reports, Amazon has decided to reverse this feature in favor of a fully cloud-dependent model, which aligns with the upcoming integration of Alexa+.

In an update, Amazon clarified that while the local processing of voice requests will end, certain key functions, such as wake word detection and visual ID, will still occur on-device. The company also pointed out that the “Do Not Send Voice Recordings” option was only available on select Echo devices—such as the Echo Dot (4th Gen), Echo Show 10, and Echo Show 15—and was used by a small group of customers. Once the local processing feature is discontinued, Amazon will automatically update users’ privacy settings to delete voice recordings after they have been processed in the cloud.

This shift to cloud-based processing reflects Amazon’s evolving strategy for Alexa and its commitment to enhancing the functionality of its virtual assistant. While the move may raise privacy concerns for some users, Amazon has made efforts to ensure that voice recordings are deleted promptly, reinforcing its focus on maintaining user trust. With the rollout of Alexa+, the company aims to deliver a more sophisticated and efficient AI experience, but it remains to be seen how users will react to the change in privacy settings.