Katalon builds AI-powered code testing companion on Amazon Nova

Many developers struggle to validate code quality quickly and effectively. Katalon identified this growing challenge and saw an opportunity to build an AI companion to address it. Around the world, software engineers—especially those without access to enterprise-level tools—often lack the expertise or resources to test code efficiently. They rely on manual tooling or costly third parties to speed up delivery, which can lead to bugs, broken links, and performance issues reaching production. “We felt this AI use case would truly make a difference in the lives of developers, especially the large community of ‘non-developers’ globally,” says Huy Tieu, senior product manager of Katalon.

To create a web-based AI companion, the team looked for flexible AI and agentic AI cloud technologies that could accelerate development. The infrastructure needed to be reliable and scalable to support user growth and heavy AI workloads, while offering the flexibility to iterate quickly. Katalon also sought a provider that could collaborate closely and offer expert guidance throughout the build. “We wanted a cloud provider that could move as fast as we did, helping us turn our idea into a working product within weeks,” says Son Dao, principal software engineer of Katalon.

Using AWS generative AI solutions, the team developed Katalon Scout, an AI-powered exploratory testing agent designed to harness generative and agentic AI capabilities. It leaned into the range of different large language models (LLM) in Amazon Bedrock to optimize the balance between performance, cost, and accuracy. Additionally, the business used Amazon Nova Act (Science Preview), a new model trained for AI agents operating in a web browser, together with Amazon Bedrock AgentCore to build, deploy, and manage agents at scale.

“Two capabilities of Amazon Nova Act stand out for us,” says Son Dao. “Visible reasoning gives our developers clarity into how the model makes decisions, which helps them guide workflows more effectively. And structured data extraction is essential for validating application states in automated testing. In our benchmarks, Amazon Nova Act consistently delivers higher accuracy than other solutions—a level of reliability that makes it a strong fit for our ecosystem.”

The ease of adoption also accelerated delivery. Katalon built Scout in just one month, supported by an intuitive setup process and well-structured Python SDK documentation. The team reported consistent performance across the environments they use in development.

Son Dao adds, “The AWS prototype team was supportive and collaborative, proactively providing the best solutions and working closely with us throughout the process. This collaboration helped Katalon fully tap into the AWS technology stack, particularly Amazon Nova Act, to accelerate the launch of our new product, Katalon Scout, ensuring stability and scalability from day one.”

Launched in October 2025, Katalon Scout has transformed development testing by delivering 100 percent AI-generated test coverage with no scripts to write or maintain. In its first 30 days, it completed more than 1,500 executions and required no ongoing QA maintenance, as its AI adapts automatically to application changes. Benchmarking showed a 60 percent reduction in test durations, while automated script conversions and validations improved overall accuracy and coverage.

Developers have already identified more than 11,000 issues in their web applications. Many are using service features, such as Traffic Light Reports for instant feedback on code quality, Evolution Tracking to monitor changes across builds, and AI Fix Prompts for targeted remediation. Katalon Scout also integrates with popular coding platforms such as Replit and Lovable, as well as assistants like Kiro and Amazon Q Developer. This broad compatibility helps developers avoid vendor lock-in and workflow disruptions by providing a neutral, flexible testing tool.

Katalon Scout strengthens Katalon’s broader AI-augmented testing platform, working seamlessly with tools like TrueTest, Studio, TestOps, and TestCloud. Together, they deliver a unified environment that brings automation, analytics, and orchestration into a single workflow.