How does BearQ autonomous QA work? Your top questions answered 

Testing software at scale has always been a race against change. Then, AI-coding turned what was once a challenge into a crisis: rapid development cycles accelerated by AI have made it impossible to maintain comprehensive test coverage and catch issues before they impact users.

In SmartBear’s Closing the AI Software Quality Gap Study, 60% of software experts told us they experienced quality issues as development outpaces testing. This number will only increase as long as teams rely only on time-intensive manual testing or easily breakable scripted automation, both of which still require significant human oversight to design, maintain, and execute test suites.  

Fundamentally shifting this current approach to software testing is our new agentic QA system, SmartBear BearQ™. BearQ autonomously explores and tests your web applications, evolving QA from a static checkpoint to a living, learning system that enables you to test at AI speed and scale. 

We recently celebrated BearQ’s launch with a livestream event where we received tons of questions and interest around its functionality and capabilities. Watch or read on as we answer some of the most common questions.

How does BearQ work?  

BearQ works by deploying a system of specialized agentic “teammates” that continuously explore, test, and report on your application’s integrity. Each agent has a defined role, and together they share an application model, which gives every agent a common, always-current understanding of how your software is supposed to behave.  

The Explorer Agent continuously traverses your application to gather context and logical intent, which it adds to the application model – BearQ’s central knowledge base. It creates a living map of pages, elements, and possible sequences of functional interaction. As your application evolves, the Explorer Agent updates this model, ensuring BearQ always has a current understanding of your software. 

Once the Explorer Agent builds a knowledge base of your application, including functional areas, you and the QA Lead Agent can define focused areas where the BearQ agents target their efforts to explore, test, and report upon application integrity. The Explorer Agent identifies high-risk areas and helps you focus testing efforts where they matter most.  

The QA Lead Agent assesses required and existing test coverage for your application based on the application model. It identifies gaps, prioritizes testing needs, and assigns tasks strategically. You can ask the QA Lead to examine the coverage of a given area and expand on it or ask it to create specific workflows if it hasn’t already. This agent thinks like an experienced QA lead, making informed decisions about where testing efforts will have the most impact. 

The Tester Agent executes the testing work, carrying out the tasks the QA Lead Agent assigns. Once the Tester Agent creates a test, BearQ stores it as a draft test for further review and refinement – either by the QA Lead Agent or by your human team members. 

This collaborative approach ensures tests meet your quality standards while maintaining the efficiency of autonomous execution. Think about BearQ functioning as members of your QA team. These agents continuously work together to build, refine, and report on your application integrity, and BearQ interacts with the UI in much the same way you do –visually and with specific intent. 

What types of applications can BearQ test? 

BearQ works autonomously on publicly accessible web applications with username/password authentication or no authentication at all. Our roadmap includes support for multi-factor authentication, single sign-on, and on-premises applications. 

We’re also expanding into API testing, mobile testing, and accessibility testing. Even now, the Explorer Agent gathers API endpoint details as it checks your application’s workflows, building a comprehensive understanding that will enable deeper testing capabilities. 

Soon, we’ll add a public API and native integration with Jira, enabling BearQ to raise tickets and respond once tasks are marked complete. This creates a seamless workflow between autonomous testing and your existing development processes. Currently, you can copy a test report from BearQ, complete with screenshots and annotations, to Jira, Confluence, Zephyr, or your bug tracker of choice. 

How do BearQ and Reflect work together? 

We received questions regarding how BearQ’s features and functionality compare to another tool in our portfolio: SmartBear Reflect. BearQ and Reflect serve different but complementary purposes across SmartBear’s testing ecosystem. They operate at different levels of autonomy that, together, create confidence through what we call layered autonomy. 

BearQ and Reflect serve different but complementary purposes across SmartBear’s testing ecosystem, operating at different levels of autonomy that together create what we call layers of autonomy. 

Reflect enables AI-assisted, human-in-the-loop automation, giving teams precision and control for high-value workflows. It’s where testers design, refine, and govern the test experience. BearQ, on the other hand, is an agentic QA system that scales quality with autonomous breadth, continuously exploring your application to find issues you didn’t even know to look for. 

Together, these tools give you the best of both worlds: the deep, controlled testing of critical paths with Reflect, and the broad, autonomous exploration of edge cases and unexpected scenarios with BearQ. This layered approach ensures comprehensive coverage without overwhelming your team. 

Integration capabilities between Reflect and BearQ via API and MCP calls are in development, so you can use the tools together soon to expand your workflows’ capabilities. 

How does BearQ execute complex testing scenarios at scale?  

BearQ handles complex testing scenarios – such as data-driven tests and combinatorial cases – using autonomous AI agents that continuously explore your application, validate complete user flows, and measure UI outcomes against intended behavior. 

For example, the Explorer Agents are super-efficient at this task, crawling and generating scenarios, data, and edge cases to ensure integrity. BearQ agents have visual comprehension models built-in, always measuring the UI outcome against the intent of the test. 

BearQ excels at multi-system workflows, highly configurable systems, and visual testing. The agents understand visual elements and can validate that what users see matches what they should see, catching issues that traditional automation might miss. 

What LLMs does BearQ use?  

BearQ’s underlying AI models are integrated through SmartBear’s in-house service platform, using optimal models selected per task rather than a single model across all functions. Customers cannot bring their own models at this time. 

How does BearQ handle authentication? 

BearQ handles authentication in two places: access to BearQ itself is securely authenticated via Microsoft or Google using your work domain, and authentication from BearQ into the application being tested. 

BearQ can support applications with no authentication, such as marketing websites, as well as applications with simple username/password authorization. We do support setting information for an app with local storage or cookies – this process can bypass bot detection or multi-factor authentication (MFA), if the app under test supports that functionality.  

How can I join the BearQ early access program? 

To sign up for early access, connect with our team and choose the path that best fits your organization. Select teams will receive white-glove onboarding with the BearQ team and have influence over product direction via your feedback. 

This is your opportunity to shape the future of autonomous testing. Early access participants work directly with our product team to ensure BearQ meets the needs of real-world development environments.  

BearQ helps you achieve quality at the speed and scale that modern software development demands. Ready to see what autonomous QA can do for your team? Join the BearQ early access program here and view our BearQ announcement restream on YouTube for more information.