How Agentic AI Is Reshaping Enterprise Testing

How Agentic AI Is Reshaping Enterprise Testing

Agentic AI is rapidly transforming how engineering teams approach software testing and quality assurance. By moving beyond traditional automation toward intelligent, agent-driven systems, organizations can streamline workflows, improve efficiency, and expand access to testing. To learn more about the state of agentic AI, Design World asked Adam Auerbach of EPAM Systems to weigh in. Below is our email Q&A correspondence, edited for length and clarity.

Design World: Design engineers increasingly rely on software tools for everything from simulations to embedded systems. From a quality engineering perspective, how do you see AI (especially agent-driven approaches) changing the way engineering software is tested and validated in the product design lifecycle?

Adam Auerbach: We really view this as a maturity curve, and at EPAM, we have defined AI maturity in levels. At Level 1, teams are using a code assist tool like Copilot or Cursor. Level 2 is where you have agents, but it's augmenting the way we work today. It's the same Software Development Lifecycle [SDLC] or Product Development Lifecycle [PDLC], but you're using agents to increase efficiency. And then Level 3, it's fully agentic.

What we're seeing right now is that, at Level 1, we can use code-assist tools to improve test automation efficiency. Think about it: in the market, there are these low-code, no-code-type solutions [Copilot, Cursor, Windsurf, etc.]. We can create agents and tools to streamline the more technical work and let more people participate. In Level 2, we're creating even more efficiencies by leveraging our existing test suite and other documentation to better design tests and adopting a more consistent test strategy. And this is where we see even more significant productivity gains. And again, we can use agents to move non-technical people into more technical roles while they continue to work in the same way. Level 3, where we see a more agentic approach, is where we have a product called Agentic QA. Basically, we can eliminate the need for traditional test automation and related tools.

What we see is that everybody must go through this maturation, because if we show someone this agentic QA approach, it's so radical that it's hard to wrap your head around — where's the code and where are my test cases, etc., etc.

Level 1 and Level 2 help speed up testing today, so you can deliver on shift left, get automation into the sprint, build efficiencies, and improve how you're working. But then Level 3, the ultimate maturity, is that you have this agentic approach. It doesn't care about the normal issues people face when they're doing test automation, which opens testing up to more people and helps us, as testers, move away from just being technical experts and being very fungible to being more like domain experts.

Read the full Q&A here.

Learn how EPAM’s Quality Engineering services can help your team make the shift toward intelligent, agent-driven systems: epam.com/services/engineering/quality-engineering