Lina Zubyte ThoughtWorks

BIO: Lina Zubyte is a passionate Quality Enthusiast who loves to ask questions, test, collaborate with diverse departments and investigate issues. Lina has worked in companies of different sizes (large multi-national companies and a startup), moved countries for work and had to adapt quickly to getting out of comfort zone. Favorite parts of being a quality professional for Lina are: diving deep to complex issues which may even reveal design or algorithm flaws, using monitoring tools and analytics data to understand the impact of found issues and collaborating with the team to build a high quality product.
In her free time, Lina loves traveling and discussions with inspiring people.

Presentation: It’s Tricky: Chatbots & QA
session level: beginner

In 2016, as part of a survey run by Oracle regarding tech trends, over 80% of businesses answered that they already have or will implement a chatbot by 2020. A few years passed, the craze for chatbots is not as sky-rocketing, but the trend is still there: more and more businesses are experimenting with chatbots, especially, when it comes to customer-facing services. However, with chatbots comes a huge amount of uncertainty.

Working with a chatbot as a QA, I felt like the traditional methods of testing do not apply as well in this new area. It made me do a lot of research, experiment, and, even build my own chatbot to explore the topic more. In this session, I will share my learnings on chatbots and discuss these questions: What should we think about when we build chatbots? How can we as QAs help to ensure the quality of a chatbot? What in the end is a good quality chatbot? Join me in my talk and get ready for the future of chatbots.


  • Chatbots being still a new growing area, there is little known about the quality metrics for conversations.
  • Validation of machine learning for chatbots currently is very challenging because humans are challenging in their conversations.
  • As QAs we need to learn to let go of control and adapt to uncertainty in order to reach quality in chatbots.