The project started with a short discovery phase followed by a proof of concept (PoC) for a particular game written in Unreal Engine. During the PoC, we tested the online and matchmaking services using headless bots for a simple “dummy” game that calls the services that need testing. The bots get the response from the service, log into the servers, complete the required actions and log out. The idea was to test the solution first on one game, and then, if successful, scale it to the entire online ecosystem and all other games. Our team designed the architecture and showcased how it works, and the solution proved to be more advanced and cost-efficient than other options. 

After a successful PoC, we moved on to the next phase where the rest of the client’s shared services were pulled into the QA bot integration testing, including the games written in C++ and Unity. Leveraging its existing AWS infrastructure, EPAM developed a swarm of up to one hundred thousand QA bots with the architecture capable of supporting millions of bots. The bots connect game clients to the online infrastructure to enable integration testing at scale. Additionally, all the bots universally support Unreal Engine, Unity and C++ and can be attached to any game with minimal integration effort. The bots' architecture allows for future extension of its functionality with AI-based player behavior simulation.

In addition to providing a new solution for high-load testing across its entire gaming ecosystem, EPAM enabled full integration with Datadog dashboards for unified access to test metrics and results. This means that the client can monitor the statistics and insights from the bots through the logging data, ensuring consistency and efficiency.