ServiceNow Virtual Agent Message Preview

AI powered predictive interactive messages

TYPE

Proactive AI Powered Messages

TEAM

Engineering Team

Product Manager

Content Writer

Accessibility Team

DATE

January - March 2022

TOOLS

Figma, Miro, Zoom, Builttools (Internal Tracking Tool)

OVERVIEW

ServiceNow is a cloud computing platform that helps companies manage workflows for enterprise operation.

What is VA? Virtual Agent (VA) makes it easy for employees and customers to resolve issues fast and get what they need, when they need it with an AI powered conversational chatbot.

The aim of this project is to create an AI powered chatbot that generates personalized help messages even before they click on the chatbot, using predictive patterns.

MY ROLE

I was the lead product designer on this project, collaborating with cross-functional teams including designers, engineers, UX researchers, UX writers. I created prototypes using Figma and delivered user flows, wireframes, high-fidelity prototypes, and visuals.

Problem

How can the Virtual Agent become more effective to motivate users to engage with the chatbot?

Project Preview

Virtual Agent proactively predicting a user's intention before they open the chat

Virtual Agent opened to view full predictive message

Discovery + User Research

I discussed with product managers on the next move for improving the product. I compiled information in Miro Boards, Figma files while conducting a competitive analysis on other companies like Intercom. This project kicked off a larger research initiative for proactive engagements.

Solution

By providing a proactive message that appears in personalized relevant scenarios, the users will be more likely to engage with the chatbot. I mapped out 4 user scenarios that users would most likely perform when encountering a proactive message from the chatbot.

User Scenario 1: 1 Proactive Message

User Scenario 2: 2 Proactive Messages

User Scenario 3: Closing the proactive message

User Scenario 4: Hovering over the proactive message

Constraints

After reviewing if our ideas were executed correctly, we realized there was an issue with the micro interactions and that we didn't release what was intended. Engineering insisted we push for that version but we truly believed in advocating for the user.

Original Vision: Pulling messages in real time when each message is pushed.
Implemented Design: Pulled at certain intervals, only getting the last message.

Click play to view interactive design

Next Steps

Micro-interactions are so important. The user could get frustrated and leave the chatbot completely, making the whole experience not useful.

We are hoping to implement the correct version in the next release.

For a deeper dive, please contact me!

Email me to learn more!