Designing an AI Chatbot

A small project I had the chance to work on at Rev was the visual design and user experience of a chatbot to be used in the Transcript Editor. The feature started with a hackathon project one of our engineers created, using ChatGPT to act as an AI assistant to help users create summaries and find other insights in their transcripts.




Lead Product Designer


Product Manager, Engineer


2 weeks for initial design and implementation

The Rev chatbot in action (live beta version)

Using AI to help users find transcript insights

The initial ask was to design a sleek UI for the beta version that fit the look and feel of Rev's branding but also felt a bit more futuristic and modern. After some brainstorming I named the initial version of the chatbot Cadmus after the legendary Phoenician hero who brought writing and the alphabet to the Greeks (a name that was unfortunately later vetoed in favor of something more generic). The engineer and I worked closely together to integrate the new design into the chatbot before we released an alpha version of the chatbot to internal users.

The product manager and I then started working together to find the best ways to help users find insight about their transcripts. We did a number of user interviews and looked at data on how our internal alpha users were using the chatbot, evaluating the most common asks as well as other things users requested that we might not have considered. We used that feedback and data to help shape the initial prompt choices that were provided to users.

The beta version of the chatbot is now live (seen in the video above) and is constantly evolving. The team is currently working on other ways to integrate AI into the transcript editor, including an automatically included summary and using AI to provide users with a more intelligent search for their transcripts.

The initial design for the chatbot (I gave it a name)
The initial design for the chatbot (I gave it a name)