AI Chatbot

Guild enables people to pay for school using the Tuition Reimbursement funding product. But reimbursement policies are complicated and nuanced, and generate a lot of questions.

We built an AI chatbot to lighten the load on member services while getting users the answers they needed.

Approach

This was a fast-moving project. We needed to:

  • work with Engineering to understand the the Amazon Lex product we'd use

  • draw on existing call center database content

  • revise all content for best conversation design practices

  • infuse the copy with branded style and tone

Click the image to enlarge.

Testing

We launched to a limited number of users and gathered key analytics. These revealed that users' responses to open-ended questions resulted in confusion.

To resolve this, we opted instead for a menu-driven approach.

Click the image to enlarge.

Challenges

It was important to nail the language and make sure our copy reflected objective policy reality.

We partnered with teammates from Member Services to review and revise copy and make sure any policy misunderstandings were clarified before we deployed.

Click the image to enlarge.

Click the image to enlarge.

Click the image to enlarge.

Outcomes

The result was an AI chatbot with a limited but powerful use-case that answered some of our members most common and pressing reimbursement-related questions.

The success of the chatbot caused leadership to prioritize applying this same approach to a wide expansion of use cases.

Click the image to enlarge.