Zoobots are a family of bots that were created specifically for video calls. At a time where people gravitate towards video calls for work and social engagements, the team decided to use machine learning to explore how bots could help facilitate and make these calls more engaging.

Users of Zoobot are people who work from home, students who are e-learning or anyone who needs to have meetings on an online platform. This was specifically designed at the time of the COVID pandemic.

When a user creates a meeting call, they will be presented with a selection of Zoobots. They can ‘hire’ as many bots as required depending on their function and even adjust the level of intervention they wish to have.

The team worked on three bots specifically so that it could run live on the class zoom call. As a result, the ‘Ideas Generator’ bot helped move conversations forward if it detected a long pause. The bot used voice recognition to translate the voice into text and processed the input through an AI model trained to generate ideas. At times, the outputs were surprisingly amusing and provocative. 

The ‘Norming’ bot gently nudged people who hadn’t contributed to the call. The ‘Icebreaker’ bot picked up on phrases and produced funny gifs to lighten the mood. Sometimes, the API generated texts were not the most accurate outputs, but the team wanted to embrace the anomalies of machine learning.