Amazon expands access to Nova AI models with new developer portal

Amazon has launched nova.amazon.com, a new developer portal that expands access to its foundation models and introduces capabilities for building web-based AI agents. The announcement, made on March 31, 2025, just two days ago, marks a significant step in Amazon's strategy to democratize access to its artificial intelligence technologies.

The technology giant unveiled its Nova family of foundation models at the AWS re:Invent conference in December 2024, initially making them available only to AWS customers and internal Amazon teams. These models, which include Nova Micro, Lite, Pro, Canvas, and Reel, are designed for various applications ranging from text generation to creating images and videos.

"Nova.amazon.com puts the power of Amazon's frontier intelligence into the hands of every developer and tech enthusiast, making it easier than ever to explore the capabilities of Amazon Nova," said Rohit Prasad, SVP of Amazon Artificial General Intelligence. "We've created this experience to inspire builders, so that they can quickly test their ideas with Nova models, and then implement them at scale in Amazon Bedrock. It is an exciting step forward for rapid exploration with AI, including bleeding-edge capabilities such as the Nova Act SDK for building agents that take actions on the web. We're excited to see what they build and to hear their useful feedback."

The portal is currently accessible only to U.S.-based customers with an Amazon account. Interested users can visit nova.amazon.com to begin exploring the models or download the Amazon Nova Act SDK to start building agents. The company has indicated that broader geographic availability may follow, though no specific timeline has been provided.

Developers can access three primary text models through the platform, each with distinct capabilities and performance characteristics. Nova Micro is described as a "text-only model that delivers the lowest latency responses," making it suitable for applications where speed is critical. Nova Lite extends beyond text to incorporate multimodal capabilities, processing "image, video, and text inputs" with high speed. Nova Pro, positioned as the premium offering, provides "the best combination of quality and speed for a wide range of tasks."

For visual content generation, the platform offers Nova Canvas for creating "high quality images from text and images" and Nova Reel for producing videos from similar inputs. According to Amazon, both models represent "state-of-the-art" technology in their respective domains.

A step toward web-based AI agents

Perhaps the most significant aspect of the announcement is the introduction of the Amazon Nova Act SDK, which allows developers to build agents capable of taking actions within web browsers. This represents Amazon's entry into the emerging field of autonomous AI systems that can interact with digital interfaces to complete tasks.

The Nova Act SDK, currently available as a research preview, is powered by a specialized version of Amazon's Nova model. According to the company, this tool enables developers to "break down complex workflows into atomic commands" such as search, checkout, and answering questions about screen content. The system also supports adding detailed instructions to commands and API integration.

In its announcement, Amazon explained its perspective on AI agents: "We think of agents as systems that can complete tasks and act in a range of digital and physical environments on behalf of the user. Today, such agents are still in an early stage. The Nova Act SDK is a crucial step forward, toward building reliable agents by enabling developers to break down complex workflows into atomic commands."

A demonstration of the technology shows Nova Act searching for apartments based on their biking distance to a train station, illustrating how the agent can interpret complex requests and navigate web interfaces to find relevant information.

Technical capabilities and positioning

The Nova family of models is positioned as delivering what Amazon terms "frontier intelligence" with "industry-leading price-performance." This suggests that the company is emphasizing not only the capability of its models but also their efficiency and cost-effectiveness, which could be particularly important for widespread adoption.

The interface of nova.amazon.com appears straightforward, with clear options for different types of content generation. For text outputs, users can click "New Session" and select their preferred model from a dropdown menu. Image generation is accessible through a "Generate Image" button, while video capabilities can be explored through a gallery that showcases the potential of Nova Reel.

Amazon has emphasized the seamless integration of these models with Amazon Bedrock, its managed service for foundation models, suggesting that experiments conducted on nova.amazon.com could be scaled up through the Bedrock platform. This positions the nova.amazon.com portal as an entry point or testing ground, with Bedrock serving as the production environment.

Market context and competitive landscape

Amazon's expansion of access to its Nova models comes amid intense competition in the generative AI market. Major technology companies and well-funded startups have been rapidly deploying increasingly capable models for text, image, and video generation, as well as multimodal systems that can process various types of input.

The introduction of agent capabilities through Nova Act is particularly noteworthy, as it addresses one of the most promising frontiers in AI development. The ability to automate complex web-based tasks could have significant implications for customer service, e-commerce, and productivity applications.

By focusing on the developer community with this release, Amazon appears to be taking a platform approach, encouraging third parties to build applications on top of its AI infrastructure rather than focusing exclusively on end-user applications. This strategy has proven effective for Amazon in other domains, most notably with AWS.

Implications for businesses and marketers

For the marketing community, Amazon's announcement carries several important implications. First, the expansion of access to generative AI tools could accelerate the adoption of AI-enhanced marketing workflows, from content creation to consumer research and interaction design.

The agent capabilities introduced with Nova Act could eventually enable more sophisticated marketing automation, potentially allowing for personalized customer journeys that span multiple digital touchpoints. This might include automated product discovery, comparison shopping, and purchase processes tailored to individual consumer preferences.

Marketers working with e-commerce platforms may need to consider how AI agents could influence consumer behavior and purchasing decisions. As these technologies mature, they could become intermediaries between brands and consumers, potentially changing how marketing messages are delivered and received.

The ability to generate high-quality images and videos at scale could also impact content creation workflows, enabling more rapid testing and iteration of visual marketing assets. This might be particularly valuable for e-commerce listings, social media content, and digital advertising campaigns.

Additionally, the multimodal capabilities of models like Nova Lite and Pro could enhance market research tools, allowing for more sophisticated analysis of consumer sentiment across text, image, and video data. This could provide marketers with richer insights into consumer preferences and behavior.

It remains to be seen how widely adopted these tools will become within the marketing community, and what new best practices might emerge as they are integrated into existing workflows. However, the trend toward more accessible and capable AI tools seems likely to continue, suggesting that marketers should begin exploring these technologies and considering their potential applications.

Future developments

While Amazon has not provided specific details about future developments for the Nova platform, the research preview status of the Nova Act SDK suggests that this technology will continue to evolve. The company's invitation for developer feedback indicates that it is taking an iterative approach to the development of these tools.

The current U.S.-only availability also points to potential geographic expansion in the future, which could make these technologies accessible to a broader range of developers and organizations globally.

As foundation models continue to advance in capability and efficiency, it is reasonable to expect that Amazon will update its Nova models accordingly, potentially introducing new specialized models for specific applications or domains.

Timeline of Amazon Nova developments

• December 2024: Amazon introduces the Nova family of foundation models at the AWS re:Invent conference, making them available to AWS customers and Amazon internal teams.
• March 31, 2025: Amazon launches nova.amazon.com, expanding access to Nova models to developers and tech enthusiasts in the United States.
• March 31, 2025: Amazon releases a research preview of the Nova Act SDK, enabling developers to build agents that can take actions in web browsers.
• April 2, 2025: The portal is operational for U.S.-based users with Amazon accounts, offering access to five models: Nova Micro, Nova Lite, Nova Pro, Nova Canvas, and Nova Reel.

The introduction of Nova Act and the expanded access to Amazon's foundation models represents a significant step in the company's artificial intelligence strategy, potentially influencing how developers, businesses, and consumers interact with AI technologies in the coming months and years.