July 28, 2025
5 min read
Amazon Web Services
Discover how Amazon Nova Act SDK enables reliable, scalable browser automation agents with AWS integrations for enterprise workflows.
Amazon Nova Act SDK (Preview): Path to Production for Browser Automation Agents
In early 2025, Amazon introduced the Amazon Nova Act SDK as a research preview to help developers build agents that reliably complete tasks in a web browser. Now, in a limited preview, customers can take their agents to production with new AWS integrations including AWS Identity and Access Management (IAM) for secure credentialing, Amazon Simple Storage Service (Amazon S3) for data storage and policy control, and the new Amazon Bedrock AgentCore Browser Tool for scalable, cloud-based browser execution. This article explores what makes the Amazon Nova Act SDK unique, how it works, and how teams across industries are already using it to automate browser-based workflows at scale.Challenges with Traditional Automated Business Workflows
Many business operations require browser interaction, such as submitting time-off requests, processing invoices, accessing vendor portals, or reviewing dashboards. Due to limited API coverage, these workflows are often manual, involving repetitive copy-pasting, multi-step flows, and navigating complex interfaces. Traditional rules-based browser automation frameworks struggle in dynamic web environments. Changes in page structure, like new form fields or dropdown options, often break brittle selectors, leading to high maintenance overhead. Moreover, these frameworks do not scale well; for example, verifying professional licenses across 50 different state websites requires building 50 unique automations. Humans adapt easily across different tools and interfaces—for instance, drafting an email in Outlook translates easily to Gmail. Large language models (LLMs), trained on millions of UI examples, offer similar pattern recognition for AI agents. These models have powered chatbots, summarization, and coding copilots by understanding language and reasoning across domains. The next phase, agentic AI, extends LLMs from understanding to execution, enabling completion of dynamic, multi-step workflows such as filling complex forms or interacting with evolving UIs at scale.Agentic AI with the Amazon Nova Act SDK
The Amazon Nova Act SDK enables building and deploying reliable browser agents powered by the Amazon Nova Act model—fine-tuned for agentic AI. Trained with reinforcement learning and extensive browser interaction data, it executes workflows step-by-step with precision. The SDK supports installation with a single command, writing agents in Python and natural language, real-time debugging, and integration into continuous integration and delivery (CI/CD) pipelines. With enterprise-grade security, observability, and infrastructure through AWS, it offers a fast, flexible path from prototype to production. The SDK integrates with the Amazon Bedrock AgentCore Browser Tool, a fast, secure, cloud-based browser enabling AI agents to interact with websites at scale. Features include virtual machine-level isolation, federated identity integration, live viewing, AWS CloudTrail logging, and session replay for troubleshooting and compliance.Benefits of the Amazon Nova Act SDK
Reliability: Build Robust Browser Automation
Developers can decompose complex workflows into atomic commands (e.g., collecting all form elements and returning required fields). The SDK allows detailed instructions (e.g., dismiss popups), API calls, and direct browser manipulation via Playwright for sensitive tasks like password entry. Python code can be interleaved for tests, breakpoints, assertions, and parallelization. Early enterprise workflows demonstrate over 90% reliability, including automated quality assurance, complex form handling, and process execution. Enhanced reasoning and recovery enable agents to adapt to UI changes and complete sequences accurately.Speed-to-Market: Prototype to Production in Days
The SDK enables rapid automation development without brittle scripts. Features include:- Git-based CI/CD pipelines
- Reusable agent modules
- Playwright fallbacks for sensitive tasks
- Threading and parallelization for speed Agents fit into existing developer workflows, enabling quick transition from experimentation to production.
- Getting Started with AWS
- What's New at AWS
- AI Agents: Capabilities, Risks, and Growing Role
- AI-Driven Crypto Trading Tools Reshape Market Strategies in 2025
- Understanding Cryptocurrency Compliance
Security: Enterprise-Grade Automation
Integration with IAM controls access to the model like other AWS services. The SDK supports macOS, Linux, Windows, and WSL2 with isolated runtime environments and encryption for sensitive data. It is designed for enterprise reliability, observability, and security.See It in Action: Streamlining Financial Decisions
In financial services, speed and accuracy in turning fragmented public data into insights are critical. The Amazon Nova Act SDK automates information gathering to support these needs.Where the Amazon Nova Act SDK Can Make an Impact
Browser-based workflows are often manual and error-prone. The SDK automates these tasks, improving accuracy and freeing teams for higher-value work. Early customers are applying it across industries:Automated Data Entry and Form Filling
Automates repetitive input across CRMs, HR tools, and finance platforms. For example, healthcare staff assist members with complex benefit applications, and public sector caseworkers re-enter household data across systems. The SDK navigates dynamic fields and popups reliably. Rackspace Technology and Alvee Health use the SDK to automate public benefits registration, expecting a 30% increase in successful registrations and up to tenfold improvement in caseload management.Customer Support Augmentation
Automates backend browser tasks across retail, travel, and SaaS support workflows, allowing agents to focus on customer interactions. Navan uses the SDK to automate payment form filling across diverse vendors, reducing repetitive tasks and scaling operations efficiently.Automating High-Stakes Administrative Workflows
Automates credential verification, identity checks, and compliance tasks across numerous third-party portals with inconsistent layouts. Automation Anywhere integrates the SDK into its Process Reasoning Engine to automate professional credential verification, enabling AI agents to reason through UI processes with human-like expertise.UX and QA Testing Across Dynamic Interfaces
Simulates real user interactions on frequently changing websites. Teams write and update tests using natural language or Python, adapting quickly without brittle selectors. Tyler Technologies automates software testing with the SDK, converting manual test plans into automated suites rapidly, saving hundreds of hours and improving product quality.What’s Next for the Amazon Nova Act SDK
Amazon is working with early customers to expand the SDK’s capabilities beyond browser workflows, applying reinforcement learning to more complex real-world tasks and deepening AWS ecosystem integration for faster prototyping and deployment.Get Started with the Amazon Nova Act SDK
Technical leaders and developers can start prototyping with the research preview at Amazon Nova Act. The Amazon AGI Lab focuses on building AI agents capable of real-world actions across digital and physical environments. Learn more at Amazon AGI Labs. Contact AWS to express interest in productionizing your agent (limited preview).About the Authors
Lori Knapp is a Principal Product Manager with Amazon Nova, leading product efforts to define how foundation models power intelligent agents across diverse real-world use cases. Her background includes Alexa voice experiences, Microsoft product strategy, and behavioral science consulting. Tara Raj is an Engineering Manager at Amazon working on Nova Act, focusing on developer experience and adoption. She has over 10 years of engineering experience, including work on Nova multimodal capabilities, Windows Subsystem for Linux, and Visual Studio Code.Resources
Originally published at Amazon Web Services Blog on Mon, 28 Jul 2025.