Talks

Let's build a simple RAG bot based on Heisenbug conference program.

This is not a talk about the so-called successful success of large companies. This is an engineering approach to implementing AI in testing for small teams where there is no extra time and budget. I'll tell you how we saved 20% of the testers' time, and share the ready-made starter package: models, scripts, and a free tool.

Errors occur at all stages of software development. The sooner we find out, the cheaper it is to fix them. At WS, testers will learn how to identify errors in requirements before the development stage. This will reduce the number of potential errors, the time and effort of the team to fix them.

LLM-as-a-Judge in AI agent testing automation — let's discuss why this practice is vital, which models to choose, why LLM-as-a-Judge will deceive you with its simplicity, and how to start writing regression automation for your AI agents.

We will tell you what problems appeared, how we learned to reproduce the defect, how we stepped on a rake, and how it all ended (or did not end).