Nina Lyow

 Banking Virtual Assistant Powered by Kore.ai + watsonx

*The contents of this case study are based on personal experiences and professional expertise. Please note that certain details and information have been omitted or altered to protect the privacy and confidentiality of individuals and organizations involved.*

Team: Sheyda Saponar, Alex Boudreaux, Jesus Olivera, and Nina Lyow

My Role and Tasks:

• Translated AI Engineers’ technical expertise + Account Lead product vision into a virtual agent for commercial banks

• Created human to AI dialogues following conversational AI best practices

• Lead video production meetings with stakeholders to review script and video

• Produced all video content including interaction design and audio

Disciplines: UX/UI design, interaction design, user research, Data and AI technical writing, Video Editing

Project Duration: 2 weeks

Deliverables:

• Video Script

• Storyboard

• High fidelity wireframes

• Sales prospecting video

Overview

Virtual agents powered by genAI provide valuable assistance to customer support centers. Especially in financial institutions, virtual agents can have advantages like answering common banking questions so that customer agents can have more time to support more difficult customer questions, and also supporting customer agents in providing accurate and timely answers to customers.

My squad within IBM Client Engineering wanted to showcase the diverse range of benefits that genAI can bring to banking virtual agents. They built and demoed a solution using a virtual agent powered by Kore.ai and watsonx but were not able to incorporate their full vision for the solution due to time and software version limitations. This is where I came in to design a new banking virtual assistant experience that incorporates 70% of the built technology and 30% art of the possible features. I also packaged the feature in a easily digestible video that our team could share with clients.

Storyboard

To align with the team on my vision for the video, I created a storyboard to show the flow of the video and included just enough visual details that depicts the illustration style, and on-screen text content that will be shown in the video.

AI Conversation Design

The virtual agent experience showcases 5 advantages of genAI

1. Natural language understanding - ability to engage in real conversations with users

2. Relevant conversation exchanges - everything said within an exchange is relevant to either the topic or previous messaging.

3. Retrieval Augmented Generation (RAG) ie. retrieve information directly from a bank’s website/policies to answer customer questions

4. Sentiment analysis ie. provide customer sentiment to call center agent so that the agent can provide an empathetic response

5. Content Generation ie. generate recommended response for call center agent based on conversation

High Fidelity Prototyping and Interaction Design

The challenge of interaction design for a continuous conversation between a customer and customer service agent was focusing on showing one conversation point of view at a time. My solution to focus on one point of view at a time was alternating a semi-opaque layer on top of a either the customer or customer service agent chat interface.

Final Video