Conversational AI Concierge

DISCLAIMER FOR CHASE CASE STUDIES

Due to prior contractual obligations, all JPMorgan Chase case studies have been recreated and purposefully anonymized to exclude any information pertaining to the company. Carol-Anne could not exit her role with any design work she produced.

Conversational AI Concierge

User Experience Design Case Study

ROLE + RESPONSIBILITIES

As the Branch Innovation Lab's Lead UX Designer, Carol-Anne was responsible for:

User Research: Conducting interviews and observations to understand customer behaviours and banking needs.

Experience Strategy and Conceptual Design: Developing the conversational AI avatar, user flows, and overall service design.

Conversation Design: Design and construction of the possible conversations and intents for the underlying designed IBM Watson machine language program

Design Lead + Team Collaboration: Partnering with senior management, product owners, artificial intelligence and machine learning engineers, and software developers to integrate biometric and conversational design technologies.

Client JPMorgan Chase Branch Innovation Labs

Project Timeline ~Six months

Team The Branch Innovation Lab, where Carol-Anne was the team's Experience Designer, collaboration partners including software engineers, the firm's Design Research team, IBM's Watson AI and machine learning engineers, and biometric technology vendors (such as Soul Machines, Idemia, Incode, and Mitek).

Problem Statement In response to evolving customer needs, the bank aimed to develop an AI-driven concierge system that could automate routine tasks, freeing up human tellers for more complex tasks while maintaining a high level of customer personalization.

PROJECT OVERVIEW

Carol-Anne Ryce-Paul, a Senior User Experience Design Lead, led the design of a conversational AI assistant integrated with a biometric authentication system for a significant financial institution. The project featured a bespoke “Soul Machines Digital Person” artificial intelligence avatar serving as an in-branch concierge. The avatar enhanced customer experiences by streamlining services such as account balance inquiries, fund transfers, and debit card replacements. The assistant recognized enrolled customers using biometrics and facilitated seamless authentication through multiple touchpoints.

This case study showcases Carol-Anne’s systems, service and experience design processes, focusing on the customer, the business strategy, design, conversation flows, and technology integration.

GOALS

The objective was to design an AI-driven, biometric-enabled system that:

Authenticates customers enrolled in the bank’s biometric program.
Provides personalized banking services with a focus on ease, security, and customer engagement.
Frees branch staff to focus on more complex customer needs.

Success criteria included:

Seamless customer authentication through facial recognition, NFC-enabled cards, and QR codes.
Successful integration with the bank’s backend for transactions and services.
Positive customer feedback on interaction with the AI concierge.

DESIGN PROCESS

Initial Research + Discovery

Carol-Anne collaborated with team members to gather data and insights about the bank’s customers, branch employees, and the technical and regulatory infrastructure. The project began with a discovery phase involving key stakeholders, engineers, and technology specialists.

Key Research Techniques

Interviews and surveys of key customer segments to understand their preferences and concerns regarding using biometrics in a financial setting. Personas such as Alex and Adrian were created to represent various customer profiles.

Observational research of how customers interacted with kiosks and human tellers provided valuable insights into pain points and opportunities for AI solutions.

Carol-Anne led further research and facilitated workshops, drawing insights from the in-branch customer ecosystem and stakeholder concerns, including:

Prospect Customers Those new to the bank navigating biometric enrollment and services.

Existing Customers Regular users are looking for efficient and familiar in-branch services.

Bank Employees Need support for high-touch, personalized services.

Branch Systems The various branch touchpoints and banking systems involved in servicing customers.

Digital Concierge Handling self-service transactions and simple banking tasks

The innovation team encountered key use cases, including biometric authentication, debit card replacement, funds transfer, and appointment scheduling, to test the project’s viability.

Customer Personas

Carol-Anne developed several personas to guide the design process. These included:

Alex A tech-savvy recent graduate managing multiple accounts and seeking personalized advice.

Adrian A retired businesswoman interested in using new technologies for secure banking.

These personas informed the service flows, ensuring the solution met diverse customer needs.

Digital Person AI Persona + Design

Carol-Anne designed a fully bespoke “Digital Person” representing a young branch banker using Soul Machines’ DNA Studio. The AI concierge was crafted to engage customers with the following:

A friendly, trustworthy demeanour. Smiling, making eye contact, and using the customer’s name.

With empathy and real-time responses, the AI concierge could adjust the tone based on customer cues and continue conversations across different touchpoints.

Conversation Design

Each service flow was paired with conversational dialogue and machine learning intents for IBM Watson, ensuring a natural interaction between the concierge assistant and the customer. For example:

For transfers, "I can help you move funds quickly. Which account would you like to transfer from?"

For authentication, "Please scan your face to confirm your identity. This will only take a moment."

This approach created a friendly, customer-focused interaction throughout the journey.

User Flow Design

Carol-Anne and the team mapped out several key user flows to ensure a seamless digital concierge experience, including:

Biometric Enrollment User Flow

Trigger A new customer approaches the kiosk and chooses biometric enrollment.

Process The concierge assistant walks the customer through face-scanning and QR code generation via the mobile app, linking their profile to the bank account.

Outcome The customer is authenticated, and their profile is created for future use.

Fund Transfer User Flow

Trigger An authenticated customer asks the concierge assistant to transfer funds.

Process The concierge assistant guides users through account selection, amount entry, and transfer review.

Outcome Funds are transferred, and a receipt is emailed.

Debit Card Replacement User Flow

Trigger A customer requests to replace a lost debit card.

Process The concierge assistant verifies the customer's identity and reasons for replacement and prompts confirmation.

Outcome A new card is issued, and delivery details are confirmed.

Sprint Conversation Dialog Flow Authentication Flow

DESIGN TOOLS + TECHNIQUES

Personas and Empathy Maps Carol-Anne crafted detailed personas using previous Lab research to understand the branch bankers and historical and continuing research to visualize ongoing customer goals and needs.

User Flow Diagrams Figma, and Adobe Illustrator were used to develop clear, visual user flows for each transaction type.

Conversational Flows The AI dialogue was structured modularly using observational research and conversations with branch bankers and customers, IBM Watson, and Figma, ensuring scalability across various services.

Visual Design + Mockups

The AI concierge design incorporated Soul Machines’ lifelike avatars with a clean, minimalist interface. Visual mockups of the kiosk interface and mobile interactions were developed using Figma, Adobe Illustrator and Photoshop.

Digital Human Concierge - Callie - Screen Example

PROTOTYPING + TESTING INSIGHTS

Conversational AI assistant testing resulted in significant improvements for the bank.

Reduced wait times Customers could handle simple tasks like fund transfers and debit card replacements through the concierge assistant, reducing the burden on human bankers.

Increased customer satisfaction Customers appreciated the ease, engagement, and personalization of the conversational assistant during the biometric authentication process.

Enhanced employee efficiency Thanks to the automation of routine tasks, branch employees could focus on high-touch, complex service requests.

KEY TESTING METRICS

Customer Wait Time Reduction
Increased Biometric Enrollment
Customer Satisfaction Scores
Employee Efficiency

TESTING CHALLENGES

Challenges arose during testing in the mock branch, and through design iterations, strategic design, and collaboration, the team continued to gather and analyze data to measure the project’s impact. The key metrics began to demonstrate the potential success of an in-branch AI concierge system.

Privacy Concerns with Biometric Authentication

Solution Carol-Anne designed the system with clear, transparent messaging to reassure customers about how their biometric data would be used and stored. She worked with legal teams to ensure compliance with privacy regulations and incorporated clear opt-in and opt-out flows within the AI’s interaction.

Ensuring Accessibility for All Customers

Solution By conducting thorough testing with a range of users, including older and less tech-savvy customers, Carol-Anne refined the AI’s interface to ensure compliance with ADA guidelines. This included but was not limited to, larger text, voice prompts, and easy-to-understand icons.

Adapting Artificial Intelligence Conversations to Varying Customer Needs

Solution The Digital Human concierge conversational AI needs to strike the right balance between offering empathetic guided support and allowing more independent users to move quickly through tasks. Carol-Anne designed flexible conversation flows, enabling users to follow step-by-step instructions or to navigate directly to their desired service needs.

STAKEHOLDER COLLABORATION

Throughout the project, Carol-Anne worked closely with various teams to ensure the design aligned with technical capabilities and business goals.

Key Stakeholder Collaboration

Branch Employees

Human and AI Collaboration: One of the main project goals was to free up branch staff from routine tasks without alienating customers who prefer human interaction. The team facilitated workshops with branch employees to understand how they understood the AI fits into their workflow. This feedback informed how the assistant concierge routed complex queries to human bankers when necessary.

Senior Management + Product Owners

Balancing Business Objectives: The Lab presented regular progress updates to senior management and product owners, ensuring that the design and service met the bank’s business objectives, including reducing wait times, improving customer satisfaction, and maintaining high-security and regulatory standards.

Engineers + Developers

Integration with Biometric Systems: The team worked alongside AI engineers to ensure smooth integration between the Soul Machines Digital Human AI concierge, the biometric authentication system, the many bank's branch systems, and IBM's machine learning. Regular design sprints helped address technical challenges, such as ensuring that facial recognition worked across the branches' different lighting conditions and that the branch's tablets, monitor speakers, and microphones worked with the various noise levels in the branch.

REFLECTIONS

Carol-Anne reflected on her experience leading the design of the AI concierge system:

The Importance of Iterative Design The project highlighted the value of early testing and iterative feedback loops, especially when designing for a diverse user base.

Balancing Innovation with Usability While the conversational AI and biometric systems were cutting-edge, Carol-Anne focused on ensuring the technology was approachable and accessible for all users, regardless of customers’ familiarity with banking tech.

Collaboration Across Teams Successful integration of multiple technologies, including biometric systems, AI, and backend banking infrastructure, required close collaboration with developers, researchers, engineers, product owners, and legal, security, and regulatory teams. Carol-Anne’s ability to facilitate these cross-functional discussions was key to the project’s success.

OTHER CHASE CASE STUDIES