AWS Outlines Agentic QA Automation Using Amazon Nova Act

Amazon Web Services has published the third installment in a blog series on using Nova Act, the company's AI agent tool, to automate software quality-assurance testing, according to the post titled "Accelerating software delivery with agentic QA automation using Amazon Nova Act – Part 3," indexed by Google News. The post continues AWS's examination of what the company calls agentic QA automation: using AI agents, rather than fixed test scripts, to check whether an application behaves as intended before a release ships.

What is "agentic QA automation"?

The phrase describes a departure from conventional automated testing, where engineers write scripts that execute a predetermined sequence of clicks and checks. An agentic approach instead gives an AI system a goal — confirm that a feature works as specified — and lets the agent decide which actions to take to test it, adjusting as the interface changes. AWS frames Nova Act as performing that role inside a QA pipeline, according to the post's headline. The specific workflows, test environments, and outcomes AWS describes in the full walkthrough were not included in the material reviewed for this report, so readers evaluating the approach should consult the original post directly.

Why does the post carry a "Part 3" label?

The numbering indicates AWS has published at least two earlier entries on the same subject before this one, consistent with how the AWS technical blog typically structures multi-part engineering guides that walk through a tool or architecture in stages. Neither the titles, publication dates, nor content of the earlier installments were available in the source material reviewed, so this account is limited to what the Part 3 headline and its indexing by Google News describe.

What problem is agentic QA automation meant to solve?

Software teams that rely on scripted test automation routinely cite the cost of maintaining those scripts: a script written against one version of an interface can break the moment a button moves or a label changes, forcing engineers to rewrite it before it can run again. Vendors pitching agentic testing tools, AWS included, argue that an AI agent capable of interpreting a task rather than following fixed steps reduces that maintenance burden. Whether Nova Act achieves that in the scenarios AWS tested, and at what accuracy or cost, is addressed in the full post rather than in its headline.

Where can engineering teams read the full walkthrough?

The original post is reachable through its Google News listing, which links to the AWS technical blog. AWS has not, based on the material reviewed for this report, issued a separate press release or regulatory filing tied to the post, and no AWS spokesperson was quoted in the available source.

What should readers do before adopting the approach?

Teams considering agentic QA tools generally benefit from testing a tool against their own application before committing to it broadly, since AI agents can behave differently depending on an interface's layout and complexity. AWS's post is pitched at engineering audiences already familiar with Nova Act and with QA automation generally, and it does not, from what is available here, include independent benchmarking against competing testing approaches.