Adam Dimla.

Project Type: Self-Started Case Study

Timeline: 2024

Brief: Discover how Conversational AI Assistants may be used in the context of personal banking.

Key Findings: It was found that users would like an AI assistant to help them with their personal banking in specific scenarios and capacities. Opting out of the feature entirely is also important to users.  

We are in the middle of a new renaissance in tech. AI models and their form-factors are changing at an unprecedented speed.

 $738bn USD by 2030 - The market share of AI is growing at an exponential rate and will skyrocket in the next few years (1).

 Samsung -  In their most recent iteration of their Galaxy line-up of phones, they’ve doubled down on the use of their AI assistant and made it a centre-stage feature.

 Rabbit R1 - Utilising a “Large Action Model” as its base, the R1 is able to use a variety of applications and services on your behalf without the use of your phone.

Large Action Models (LAMs) are the next step in how AI can help people, but how should they be implemented into personal banking?

(1) - Statista: Artificial Intelligence - Worldwide: https://www.statista.com/outlook/tmo/artificial-intelligence/worldwide

TD Clari 

An automated chat bot service by Kai (Kasisto, Inc.). The capabilities of this chat bot are:

• Basic Queries: Like most chat bots that we’re familiar with, Clari can answer questions in regards to a user’s accounts, habits, and spending.
• Tutorial Aid: If a user asks Clari for help with certain tasks, the chat bot will bring up tutorials on how to carry out the task (I.e Making an e-transfer) and basic info.
• Clari is most likely based upon a Large Language Model (LLM) or Natural Language Understanding Model 
(NLU)
  

These functions are basic, at best, and do not offer highly integrated functionality. How do we expand upon these capabilities?

introduce a more comprehensive AI Assistant into the current TD banking experience for everyday users?

User Interviews: To understand users on a more personal level in regards to AI and its impacts on daily life. Also a good chance to bench-mark their current experiences with personal banking products.

Surveys: To gather a more general and quantitative outlook on the user base and feelings about AI assistants. Can have better reach, geographically speaking, as well.

Talking to Key stakeholder: Understanding the context of why the TD app is the way it is, is key. Speaking with the team in charge of the TD app, execs, and 3rd parties involved will be helpful.

Opportunities for Improvement: This may be a given, but there are always opportunities to make something more pleasant to use.

Comprehensive Insight: What is the current userbase like? What are their pain points
and desires? How is the world affecting their current use patterns?

Accessibility Opportunities: How might the opportunities benefit people in other ways? How can they help people who are Differently Abled?

How in-depth of an LAM AI Assistant should we implement into the current TD App Experience?

How will a more capable AI assistant be received by users? Are there any reservations?

• Users felt that digital assistants are neither helpful or a hindrance. They work in certain situations and don’t in others.
• Most users are only comfortable with some tasks being done, not all.
• Most users are hesitant about their personal info being handled by an AI.
• Assistants are helpful at simplifying moderately complicated tasks.
• Misunderstanding requests, tracking info, not efficient in some cases, personal hesitancy are all reasons why people may be hesitant to using them.