In early 2025, we launched the Amazon Nova Act SDK as a analysis preview to assist builders construct brokers that reliably full duties in an internet browser. Now, we’re excited to work with clients to take their brokers to manufacturing in a restricted preview, utilizing new AWS integrations together with AWS Identification and Entry Administration (IAM) for safe credentialing, Amazon Easy Storage Service (Amazon S3) for information storage and coverage management, and the brand new Amazon Bedrock AgentCore Browser Device for scalable, cloud-based browser execution.
On this put up, we stroll by what makes the Amazon Nova Act SDK distinctive, the way it works, and the way groups throughout industries are already utilizing it to automate browser-based workflows at scale.
Challenges with conventional automated enterprise workflows
Many day-to-day enterprise operations require a browser, corresponding to submitting time-off requests, processing invoices, accessing vendor portals, or reviewing dashboards. Lack of API protection usually means workflows are executed manually: groups copy-paste information throughout tabs, observe multi-step flows, and click on by numerous interfaces to get work executed.
Conventional rules-based browser automation frameworks usually face challenges in dynamic internet environments. Groups can spend extra time on ongoing upkeep than on constructing new automations, as a result of modifications in web page construction (for instance, newly added kind fields or dropdown choices) break brittle selectors. Most significantly, these frameworks are tough to scale. If one use case is carried out on 50 totally different websites (for instance, skilled license verification on state web sites), groups should construct 50 site-specific automations, as a result of rules-based frameworks don’t generalize.
As people, our potential to carry out duties adapts throughout totally different instruments and interfaces. For instance, as soon as you know the way to draft an e-mail in Outlook, you’ll be able to simply do the identical in Gmail—even should you’ve by no means used it earlier than. Massive language fashions (LLMs), skilled on tens of millions of examples of UIs, supply the potential to create an analogous kind of sample recognition for AI brokers. They’ve introduced us this far—powering chat, summarization, coding copilots, and extra—by deciphering language, following directions, and reasoning throughout domains. Now, we’re getting into the subsequent section of generative AI: one centered on motion. Agentic AI builds on the muse of LLMs to maneuver from understanding to execution. These techniques are designed to finish dynamic, multi-step workflows—like filling out complicated varieties, interacting with evolving UI parts, or performing real-world enterprise duties at scale. Agentic AI doesn’t change LLMs—it extends them, unlocking new automation capabilities that carry us nearer to actual activity completion in enterprise environments.
Agentic AI with the Amazon Nova Act SDK
With the Amazon Nova Act SDK, you’ll be able to construct and deploy dependable browser brokers powered by the Amazon Nova Act mannequin—purpose-built and fine-tuned for agentic AI. Educated with reinforcement studying and in depth in-domain browser interplay information, it executes step-by-step workflows with precision. With this newest model, we’ve prolonged these capabilities with AWS integrations so you’ll be able to take your brokers from prototype to manufacturing. You’ll be able to set up the SDK with a single command, write brokers in Python and pure language, debug in actual time, and combine instantly into steady integration and supply (CI/CD) pipelines. With enterprise-grade safety, observability, and infrastructure now accessible by AWS, the Amazon Nova Act SDK supplies a quick, versatile path for groups trying to construct brokers that act—and ship—at scale. You need to use the Amazon Nova Act SDK to automate real-world workflows the place conventional scripts or general-purpose fashions aren’t dependable or scalable sufficient. You’ll be able to set up it with a single command, write brokers utilizing a mixture of Python and pure language, debug whereas the workflow runs, and deploy by CI/CD pipelines.
The Amazon Nova Act SDK additionally integrates with the brand new Amazon Bedrock AgentCore Browser Device—a quick, safe, cloud-based browser that allows AI brokers to work together with web sites at scale. It contains enterprise-grade safety features, together with digital machine-level isolation and federated identification integration. The software gives built-in observability by stay viewing, AWS CloudTrail logging, and session replay to troubleshoot, keep high quality, and help compliance.
Advantages of the Amazon Nova Act SDK
The Amazon Nova Act SDK is dependable, quick to deploy, and constructed for safe, large-scale browser automation use circumstances. On this part, we talk about a number of the advantages of the SDK in additional element.
Reliability: Construct strong browser automation with excessive accuracy and repeatability
With the Amazon Nova Act SDK, builders can break down complicated workflows into dependable atomic instructions (for instance, acquire all kind parts of a webpage and return a string with all required fields of the shape). It helps the addition of detailed directions to refine these instructions when wanted (for instance, dismiss any popup banners), the flexibility to name APIs, and the choice to alternate direct browser manipulation by Playwright to enhance reliability (for instance, for getting into passwords). Builders can interleave Python code—corresponding to assessments, breakpoints, assertions, or thread swimming pools for parallelization—to optimize efficiency, particularly as a result of even the quickest brokers are constrained by webpage load occasions. With this newest model, the Amazon Nova Act SDK is already demonstrating over 90% reliability throughout early enterprise workflows, together with automated high quality assurance, complicated kind dealing with, and course of execution. Enhancements to reasoning and restoration assist brokers adapt to altering UIs and full complicated sequences persistently and precisely.
Pace-to-market: Transfer from prototype to manufacturing in days—not weeks
The Amazon Nova Act SDK is designed that will help you construct automation shortly, with out counting on brittle scripts. You’ll be able to set up the SDK with a single command. You’ll be able to outline brokers utilizing Python, pure language, or each. You’ll be able to debug flows whereas they run, examine the DOM, pause between steps, and iterate quickly. The Amazon Nova Act SDK helps the next options:
- Git-based CI/CD pipelines
- Reusable agent modules
- Playwright fallbacks for delicate duties (for instance, password enter)
- Threading and parallelization for quick execution
You don’t have to vary your infrastructure or rebuild your inner instruments. Brokers constructed with Amazon Nova Act match into present dev workflows and can help you transfer from experimentation to manufacturing shortly.
Safety: Deploy automations you’ll be able to belief—powered by AWS
The Amazon Nova Act SDK integrates with IAM for entry management, and entry to the mannequin is managed similar to entry from different AWS companies. It helps execution on macOS, Linux, Home windows, and WSL2. Runtime environments are remoted, and encryption is supported for delicate inputs and outputs. The Amazon Nova Act SDK was designed to work inside enterprise environments—with the reliability, observability, and safety that manufacturing techniques require.
See it in motion: Automating info gathering to assist streamline monetary choices
In monetary companies—particularly funding banking, M&A advisory, and strategic analysis—success usually relies on how briskly and precisely groups can flip fragmented public information into actionable perception. The next demo exhibits the Amazon Nova Act SDK in motion.
The place the Amazon Nova Act SDK could make an impression
Browser-based workflows are widespread in as we speak’s companies, but many stay handbook, repetitive, and vulnerable to error. The Amazon Nova Act SDK helps organizations automate these duties, liberating up groups to concentrate on higher-value work, enhance accuracy, and scale back operational delays. Its reliability makes it a match throughout industries and use circumstances. On this part, we offer some examples of what our early clients are constructing.
Automated information entry and kind filling
The Amazon Nova Act SDK reduces repetitive handbook enter throughout web-based techniques—like CRMs, HR instruments, and finance platforms—by automating kind submissions, uploads, and updates. In healthcare, employees help members with complicated, state-specific profit functions. Public sector caseworkers additionally re-enter family information throughout a number of techniques. The Amazon Nova Act SDK handles these dynamic flows reliably—navigating shifting fields, dropdowns, and popups with out brittle scripts or customized code.
Rackspace Expertise, a number one hybrid and AI options supplier, is working with Alvee Well being to robotically register members for public advantages utilizing the Amazon Nova Act SDK. “Many registration varieties for public applications are lengthy and complicated, so members usually don’t apply for the assistance they want,” stated Nicole Prepare dinner, CEO at Alvee. “With the Amazon Nova Act SDK and harnessing info already in Alvee’s system, we’re not simply simplifying paperwork—we’re serving to present well timed, correct entry to the sources that help more healthy lives. We count on this innovation to extend profitable profit registrations by 30%, and enhance total case load by as much as tenfold, permitting healthcare suppliers to focus extra on affected person care and fewer on administration. It is a prime instance of how AI can be utilized to help well-being and enhance total well being for communities.”
Buyer help augmentation
Buyer help groups throughout retail, journey, and software program as a service (SaaS) usually transfer between inner instruments and third-party portals to resolve tickets. For instance, a retail affiliate would possibly submit a return on a companion web site. A journey agent would possibly log in to an airline dashboard to request compensation. A help rep would possibly reset a license key in a buyer admin console. The Amazon Nova Act SDK automates these browser-based duties, serving to brokers keep targeted on buyer conversations whereas backend steps are executed reliably and at velocity.
Navan, a number one journey and expense administration platform, makes use of the Amazon Nova Act SDK to simplify its journey brokers’ workflows by automating how they fill out fee varieties throughout a variety of distributors.
Yuval Refua, SVP of Product, stated, “Including the Amazon Nova Act SDK to our brokers’ workflows has helped us scale back repetitive duties—a vital step in scaling our operations to serve extra clients. We tried different laptop use instruments, and Amazon Nova Act’s reliability and adaptability enabled a single script work throughout various fee varieties from a variety of resort manufacturers. We’re now increasing this automation to cowl much more distributors, which we count on will improve our operational capability and assist us meet rising buyer demand extra effectively.”
Automating high-stakes administrative workflows
Credential verification, identification checks, and different compliance-heavy duties usually contain navigating lots of of third-party portals with inconsistent layouts. The Amazon Nova Act SDK makes it doable to automate these workflows with excessive accuracy, flexibility, and full management—serving to groups scale whereas sustaining precision.
Automation Anyplace, a worldwide chief in Agentic Course of Automation (APA), is increasing its automation capabilities by the Amazon Nova Act SDK, beginning with skilled credential verification—a high-stakes, repetitive activity that’s important for compliance, member security, and day-to-day operations.
“By deeply integrating the Amazon Nova Act SDK into our Course of Reasoning Engine (PRE), we’ve unlocked a serious leap ahead in laptop use for enterprise automation,” stated Adi Kuruganti, Chief Product Officer at Automation Anyplace. “Our goal-oriented AI brokers don’t simply mimic clicks, they purpose by UI-based processes in actual time, navigating complicated web sites with human-like experience. This opens the door to automating beforehand out-of-reach use circumstances like healthcare program enrollment testing, the place accuracy and scale are important.”
UX and QA testing throughout dynamic interfaces
UX and QA testing usually entails simulating actual consumer interactions on incessantly altering web sites—particularly in sectors like banking, insurance coverage, and authorities. With the Amazon Nova Act SDK, groups can write and replace assessments utilizing pure language or Python, adapting shortly to UI modifications with out brittle selectors or handbook rewrites.
Tyler Applied sciences, a number one supplier of built-in software program and know-how companies to the general public sector, is utilizing the Amazon Nova Act SDK to automate software program testing and enhance the reliability of its releases. “Amazon Nova Act’s natural-language interface lets us convert our handbook take a look at plans into automated suites in minutes—with out writing a single line of code, saving us lots of of hours whereas increasing take a look at protection and growing product high quality.” stated Franklin Williams, President of Information & Insights at Tyler Applied sciences. “We’re now trying to increase the usage of Nova Act SDK throughout our portfolio.”
What’s subsequent for the Amazon Nova Act SDK
We’re working carefully with early AWS clients to tell our roadmap. Though as we speak’s focus is on browser-based workflows, the Amazon Nova Act SDK is a part of a broader effort to construct brokers that may function reliably throughout various environments. We’re persevering with to increase the mannequin’s attain past the net, making use of reinforcement studying to extra complicated, real-world duties. We’re additionally deepening integration throughout the AWS ecosystem to assist builders transfer sooner—from prototyping to deploying safe, scalable agentic techniques.
Get began with the Amazon Nova Act SDK
In the event you’re a technical chief or developer and need to begin prototyping with the analysis preview of the Amazon Nova Act SDK, go to Amazon Nova Act. You’re going to get entry to early tooling designed for dependable, step-by-step browser automation—constructed for real-world workflows, not simply demos.
The Amazon AGI Lab is Amazon’s utilized analysis group targeted on constructing helpful AI brokers that may take real-world actions in digital and bodily environments. Their work spans LLMs, reinforcement studying, world modeling, and extra. To be taught extra and sustain with their newest improvements, go to Amazon AGI Labs.
Contact us to precise curiosity in working with us to productionize your agent (restricted preview).
Concerning the authors
Lori Knapp
is a Principal Product Supervisor with Amazon Nova. She leads product efforts to outline how basis fashions can energy clever brokers throughout various real-world use circumstances. Previous to this function, Lori’s expertise spanned scaling adaptive voice experiences at Alexa, product technique at Microsoft, and behavioral science consulting. Outdoors of labor, she enjoys exploring new cities, internet hosting dinner events, and fixing crossword puzzles.
Tara Raj is an Engineering Supervisor at Amazon engaged on Nova Act. In her present function she’s targeted on developer expertise, from constructing nova.amazon.com/act to the Nova Act SDK with the software program engineers on her group to driving adoption of Amazon Nova Act together with her options architect group. Tara has over 10 years of expertise in engineering roles bringing merchandise from imaginative and prescient to launch together with Nova multimodal capabilities at Amazon and the Home windows Subsystem for Linux and Visible Studio Code at Microsoft. Outdoors of labor yow will discover her touring, dancing, and making an attempt new eating places.
