The Inter-American Development Bank created the Inter-American Institute for Economic and Social Development (INDES) in 1994. Its mission is to promote knowledge and learning about economic and social development matters in Latin America and the Caribbean region. INDES offers Massive Open Online Courses (MOOC) on the edX platform through the idbX Program.
XiBOT - a chat bot that supports students navigating through the edX platform and the IDBx courses.
INDES - IDBx Program @ Inter-American Development Bank
Lead product designer
Development and launch of the first chat bot version: November 2019 - September 2020
One of the main benefits of online courses is that users can consume them at their own pace. This is especially useful for users with limited time availability, as they can learn at any time, anywhere.
However, whenever users needs assistance, they may lose motivation, sign off and even drop out. In the absence of a 24/7 helpdesk, it may take several hours before the support team can answer a user question.
Clear instructions, a clean course design, student guides and FAQ pages, all can help to make essential information availavble. Nevertheless, users prefer to spend time reviewing the course content rather then researching course dates, download buttons or how to redeem a coupon.
User research relied on primary data stored in Service Now (a workflow management platform). When users submit a question to the support team via an online form, Service Now creates a ticket and assigns it to a specific person based on topic. The analysis of user queries revealed a set of frequent topics that informed the design solution.
The team decided to develop a virtual assistant (chat bot) that would allow users to receive answers in real time. The chat bot is easily accessible through a chat icon located at the bottom right corner of every page of the course. The team developed the chat bot in IBM Watson and integrated it in the course through an HTML code linked to the Watson API.
The team received some training on IBM Watson and developed a prototype that was tested for funcionality. Then the team developed a work methodology.
Data is clustered per topic and duplicates are deprecated.
Personas are developed using Trello based on the data available in the yearly user survey. In each card the team developed a user story.
The team developed each persona card into a user story.
Intents, entities and conversational flow is created in IBM Watson.
The Watson assistant preview is used to test the chatbot in two phases:
The chat bot was deployed in one course, then in three courses. Currently, the team is evaluating the results before deciding to scale it up.
After an initial deployment with the name ''Ximena'', it became clear that some users believed they were talking to a real person. Therefore, the team decided to change the name to ''XiBOT'' and fine-tune the welcome message.