AI Agents | Conversational AI Design | Conversational Interface

Reimagined and Designed an Interactive Digital Avatar Chat Experience for an Experimental Drug Trials.

MY ROLE

Conversational Design Lead

Product Designer

TEAM

Solution Delivery Lead

Digital Client Partners

Business Team

Dev Team

CX Team

DURATION

13 Months

TOOLS

Figma

Voiceflow

Kore.ai

Soul Machines

Jira

IMPACT

Shipped the product and is currently being used at Pfizer's clinical trial location. Since Jan 2025, patient engagement for medical clinical trial increased by 10%.

UNDER NDA

I can’t share full details of this project due to an NDA, but I redesigned the dashboard in a way that ensures I don’t break NDA while still showcasing my work.

THE PROBLEM

How might we seamlessly create a natural, human-like digital avatar experience that helps patients onboard and ask questions about clinical trials with precision and compassion?

THE PROBLEM

How might we seamlessly create a natural, human-like digital avatar experience that helps patients onboard and ask questions about clinical trials with precision and compassion?

FINAL SOLUTION

Designed a voice-based conversational interface named 'April' who acts as a nurse navigator, guiding patients through the e-consent form in a clinical trial and answering questions about the trial and experimental drug.

Digital Assistant "April" guides users through all the sections for the form.

April introduces herself and walks patients through the informed consent form, making sure they understand and are following along.

Patients can interrupt April to as any question they might have about the form.

April can answer any questions anytime, but patients need to click the hand button to interrupt.

April provides human like gesturing, empathy and dynamic responses.

Since clinical trials can be sensitive and emotional, the avatar needs to be flexible, understanding, and provide comfort to patients.

Patients can flag a section for site staff to revisit after onboarding.

Sometimes, the digital assistant may not fully clarify certain information. If that happens, patients can flag a section for a site staff member to explain.

Patients can ask April to review or repeat any section.

April can repeat or review any section of the ICD if patients need more information. They just need to say "repeat" and the section to review.

Patients can pick up where they left off after a session timeout.

If a patient is inactive for 3 minutes, they will be logged out, but they can always rejoin and continue where they left off.

AGILE PROCESS & STRATEGY

This project followed an iterative approach, with each stage acting as an Agile sprint. During each sprint, business users could explore and get familiar with the work-in-progress product.

Pre-Discovery

In this phase, we meet with client partners to discuss requirements. The Customer Experience Team conducts user research with patients, while my team performs regular feasibility checks.

Discovery and Implementation

We conducted a series of workshops to align with customers on conversational design and requirements, including finalizing and scoping them. We conducted periodic review & demo sessions with stakeholders.

Testing

Integrated bot to stage environment and conducted an end-to-end system testing.

User Acceptance Testing

We conducted 5 participatory design sessions with 3 Non-technical founders and 2 technical founders. This was to get a deeper understanding of how the AI copilot can help with better collaboration and iterations.

Deploy & Hypercare

SCOPING REQUIREMENTS

Collaborated with business and dev team to adjust requirements based on timeline, feasibility and enhancing experience.

CONVERSATIONAL DESIGN

The main goal was to improve the conversational flow.

I started by studying how nurse navigators interact with patients and mapped these interactions in Voiceflow. The second goal was to test the digital assistant's voice to ensure it sounded natural, approachable, experienced and supportive.

THE FINAL SCREENS

After bringing everything together, we designed a 0-1 AI Copilot to help founders collaborate, brainstorm, and refine their product more effectively.

REFLECTION

This was my first experience as a sole designer owning the end-to-end product process, and I learned the crucial importance of feedback and cross-functional collaboration.

Embrace Feedback: More feedback leads to a better end product. Different perspectives can greatly improve the user experience.
Cross-Functional Collaboration: Working with the business and legal teams improved my presentation and storytelling skills.
Importance of Storytelling: Explaining my design approach to various departments enhanced my storytelling abilities in conversational design.
Attention to Detail: Every design decision needed to be backed by quality data, highlighting the importance of detail in creating a successful product.

Embrace Feedback: More feedback leads to a better end product. Different perspectives can greatly improve the user experience.
Cross-Functional Collaboration: Working with the business and legal teams improved my presentation and storytelling skills.
Importance of Storytelling: Explaining my design approach to various departments enhanced my storytelling abilities in conversational design.
Attention to Detail: Every design decision needed to be backed by quality data, highlighting the importance of detail in creating a successful product.