First Orion Cuts QA Bottlenecks by Replacing Selenium Scripts with Amazon Nova Act Agents
Branded communications company First Orion adopted Amazon Nova Act as a pre-release partner in March 2025 to replace fragile Selenium and Playwright test scripts with natural-language QA automation. The shift let QA analysts author tests directly without waiting on automation engineers to translate test cases into code.
The Problem: QA Couldn't Keep Pace With Shipping Velocity
First Orion, a branded communications company whose call-identification and anti-spam solutions reach hundreds of millions of phone calls across carriers in the US, Canada, UK, and Germany, has adopted Amazon Nova Act to automate quality assurance testing across its web applications, according to a joint post from AWS and First Orion.
The company, founded in 2008 and now employing over 300 people across offices in North Little Rock, Seattle, London, and Dubai, serves major carriers including T-Mobile, Verizon, AT&T, Vodafone, and Deutsche Telekom. As First Orion expanded from a monolithic web portal into a decentralized, cell-based architecture — and pushed further into the small and medium-business market — the number of web applications and device/browser combinations requiring testing multiplied faster than its QA team could handle.
Three specific failures compounded: regression testing wasn't self-service due to dependency-heavy test cases; new features couldn't be tested until they were fully built and deployed, since automation engineers needed DOM selectors that didn't exist yet; and existing Selenium/Playwright scripts broke constantly as element IDs and classes changed between sprints. Adding headcount didn't fix the underlying fragility of selector-based automation.
The Shift to Natural-Language Test Automation
First Orion became a pre-release adopter of Amazon Nova Act in March 2025, after its AWS account team introduced the tool, and reported value ahead of the product's general availability. Nova Act lets QA analysts describe test steps in plain English — for example, "Log into the portal, navigate to billing, and verify the invoice total" — rather than writing and maintaining selector-based code.
According to First Orion, the agent reasons about the current UI state using visible labels, layout, and context rather than fixed CSS selectors, meaning instructions like "click the Submit button" continue to work even after developers rename underlying classes. The company says Nova Act also handles dynamic content, dismisses pop-ups, and recovers from errors autonomously, and that developers can interleave Python code, assertions, and parallelization directly with Nova Act commands.
First Orion built a custom system around the Nova Act SDK: a React-based Test Case Authoring UI lets QA analysts write and validate test cases in plain English, with a templating engine generating realistic dynamic data (phone numbers, emails, business names) for each run. Test collections are stored as JSON in Amazon S3, and execution runs on Amazon Bedrock AgentCore Browser, which the company says removed the need to manage its own browser infrastructure for provisioning, session recording, and parallel test execution.
Reported Outcomes
According to First Orion, the switch eliminated three structural bottlenecks: QA analysts can now author tests directly without waiting for automation engineers to translate requirements into code; tests survive UI changes since they don't depend on fixed selectors; and the company no longer needs to maintain its own browser test infrastructure. The source post does not disclose quantified metrics such as reduction in regression time, defect escape rate, or release-cycle acceleration — these are described qualitatively rather than with hard figures.
What This Means
This is a customer case study, not a new model or product launch — Amazon Nova Act itself was previously announced as an AI agent for browser automation. The significance here is architectural: it demonstrates a live production use case where natural-language, vision-based UI agents replace selector-dependent test frameworks like Selenium and Playwright, a pain point familiar to nearly every engineering org running continuous deployment. The claimed benefit — QA analysts writing executable tests directly, without an intermediate coding step — could meaningfully compress the last-mile bottleneck between feature completion and release if the resilience-to-UI-change claims hold up at scale beyond this one deployment. Enterprises evaluating agentic QA tools should note that this account comes from a company co-authoring the promotional post with AWS, and independent benchmarking of Nova Act's reliability against traditional frameworks is not yet publicly available.
Related Articles
AWS Publishes Reference Architecture for Deploying Anthropic's Claude Apps Gateway at Enterprise Scale
AWS published a production reference architecture for deploying Anthropic's Claude apps gateway, a self-hosted governance layer that sits between Claude Code, Claude Desktop, and Amazon Bedrock or Claude Platform on AWS. The deployment pattern centralizes SSO authentication, model access policy, and spend controls for enterprise rollouts.
Anthropic Makes Claude Code's Auto Mode Default for Pro, Max, and Team Users on August 14
Anthropic will make Claude Code's auto mode the default for Pro, Max, and Team accounts starting August 14, reducing step-by-step approval prompts. The company cites a study of 1,053 testers showing auto mode caught 89% of harmful actions versus 13.6% for manual review.
Amazon, Cursor, Microsoft, OpenAI, and Vercel Launch Agent Plugins, a Shared Packaging Standard for AI Agent Extensions
Amazon, Cursor, Microsoft, OpenAI, and Vercel have released Agent Plugins, an open standard defining a single package format for AI agent extensions. Version 1.0.0 covers Agent Skills and MCP servers, but leaves marketplaces, permissions, and runtime out of scope.
Mistral Launches Regional Inference Endpoints, Opens Platform to Third-Party Models, Targets 1GW of European Compute by
Mistral AI has made its Regional Endpoints generally available, letting customers choose EU or US inference, while opening its platform to third-party open models starting with Z.ai's GLM-5.2. The company also announced a coalition of European enterprises committing to long-term compute capacity, targeting up to 1GW by 2030.
Comments
Loading...