GeorgeBOT
Your friendly global-neighbourhood doctor, in a lightweight chatbot for Messenger and Slack.
Role
UX/UI designer
Tools
Sketch, Framer, Firebase
Timeline
One week, August 2016
Results
Concept designs and copy writing for Slack and Messenger bots, and the creds to do an Intro to Framer tutorial for the BlueDot Design team. 🙌
At BlueDot, we had another existing MVP called GeorgeBOT, which was a Slack chatbot translated from their consumer app on infectious diseases—George. Instead of showing disease risks per city, GeorgeBOT provided diagnoses of tropical infectious diseases. The algorithm existed and was translated into a bot in order for the clinical team to check their database of symptom-disease relationships.

"Instead of showing disease risks per city, GeorgeBOT provided diagnoses of tropical infectious diseases. "

GeorgeBOT's new iteration comes from the user testing results found from Wyatt. My role was to iterate on the new design of Wyatt, integrating GeorgeBOT's capabilties, while focusing to create a lightweight consumer "app" on a chat botplatform.

From bot identity, to Slack and Messenger commands, to every word of the copy, I iterated and designed GeorgeBOT 2.0.

Preview of the final design of GeorgeBOT on Messenger, prototyped using Framer and Firebase.

Conversational UI
Amongst many apps that have to gather a user's personal information, mHealth apps fly solo in the club trying to get your number—mHealth apps have no wingmen, no mutuals—and yet needs the most information in your first meeting.

BD style guide
Preview of GeorgeBOT's identity and conversational guidelines.

Based on the competitive analysis conducted for direct and indirect competitors for medical diagnostic apps and services in Wyatt, integration with the various workflows around getting treatment (such as finding readings on conditions, getting a diagnosis, keeping your records, and connecting you to a doctor) in chat-form diagnoses or doctor Q&As were found to be the best ways to gather data from the user.

"Chat-form diagnoses or doctor Q&As were found to be the best ways to gather data from the user."

It showed that the more seamless the interaction was with gathering the data from the user about personal matters such as their own health, showed better returns in terms of the users' reactions to how they perceived the quality of the answers they received.

"2.5 billion people have at least one messaging app installed, and in a couple of years, this number is expected to skyrocket to 3.6 billion."

A chatbot is better at staying relevant than a mobile app. Based on recent news and trends, the rise of chatbots is apparent. For example, the Economist reported that 2.5 billion people have at least one messaging app installed, and in a couple of years, this number is expected to skyrocket to 3.6 billion. Text-based services are convenient, reliable, and can be abandoned and picked up with little to no consequence. Most importantly, people won't have to leave their most frequently used apps.

Chatbot design

GeorgeBOT was then put back into the surface to be redeveloped. Georgebot started as a verification tool for BlueDot's symptom data, put into a chat bot in Slack that took yes or no inputs.

Georgebot
Sample conversation with GeorgeBOT 1.0 implemented in Slack.

The new vision for GeorgeBOT is to become a healthcare chatbot, with a focus on travel-related infectious diseases. Added to the diagnostic feature are features on news, health services, and travel preparations.

"The new vision for GeorgeBOT is to become a healthcare chatbot."

Thus, the game plan for GeorgeBOT 2.0 was to first increase awareness, then streamline the to-do's. BlueDot's consumer products have always had the underlying goal of increasing awareness on infectious diseases. Travel-related diseases becoming a public health concern, the solution we strive for is a proactive one.

BD sketch
Inital notes done from subject-matter expert interviews.

Aligning with the stakeholder's needs, GeorgeBOT had to be a balance of a relevant and fun consumer product, while standing as a provider of niche medical expertise on infectious diseases. Research with subject matter experts were done to recreate the feel of a doctor's visit into a bot, and see if this was an appropriate approach to a health chatbot. Moreover, various chatbots were tested and reviewed, such as Poncho the Weather cat (which I still use to this day).

"GeorgeBOT had to be a balance of a relevant, fun, consumer product, while standing as a provider of niche medical expertise on infectious diseases."

BD news
A concept design of how GeorgeBOT 2.0 will deliver news to a user in Slack.
News was put in with inspiration from the CNN Messenger bot, except this GeorgeBOT would pull from a news API and filter out keywords depending on what the user wants to know about (disease or location-wise). This appears to be the interaction with the lowest consequence and hence, would be easy to try and maybe even adopt. Starting with increasing awareness with a low consequence action, I moved on to the more streamlined processes such as travel preparation, finding health services nearby, and symptom checking.
Results

When designing chat bots, copy writing is the design. Although different messaging applications give you different degrees of freedom on how to obtain or provide information to a user through a bot, the meat if the interaction is of course, in the text. Text properties such as being too long, timing and delay, chunking, use of emojis, and providing sample replies—are just some of the may nuances I encountered when writing the logic for GeorgeBOT.

BD messenger
Sample conversation flow designed for Messenger.
This was my first project on Framer, where I aimed to prototype the Sketch files of all the possible responses and predetermined replies, while accessing and posting data from and to Firebase for a more dynamic prototype. The data inside Firebase included sample JSON from BlueDot's George API.
Previous project / Wyatt
Next project / Wealthsimple