A T O M

A GenAI-powered chatbot that simplifies medical documents

Led to a $2M client deal and generated demand from 2 Fortune 500 Healthcare clients

Due to NDA purposes not all interfaces are shown.

A B O U T T H E C H A T B O T

ATOM is an AI-powered chatbot that retrieves and synthesizes complex medical documents to answer inquiries related to Clinical Development, Product Sales, and Pharmaceutical Labeling.

This solution delivers concise insights to help healthcare professionals work more efficiently.

T H E P R O B L E M

When consulting for healthcare and pharmaceutical clients, IBM Consultants discovered

reviewing medical documents can take hours, delaying

tasks like product development.

T H E S O L U T I O N

To resolve this, IBM's Healthcare Client Executive, decided to create a

GenAI chatbot that converts complex documents into simple insights.

S T A K E H O L D E R A L I G N M E N T

Once the Client Executive aligned the solution with the team, they collaborated with the AI Strategist and Data Scientist to shape the solution architecture.

From there, the focus shifted to transforming AI capabilities into a user-friendly experience.

Before diving into the design, I met with the Product Manager to understand key requirements, ensuring the user interfaces of the chatbot would align with user needs, business goals, and technical requirements.

K E Y P R O D U C T R E Q U I R E M E N T S

Due to the tight timeline, the chatbot must be designed using Appian

(a low-code application development platform)

The app must clearly convey its product offerings and healthcare knowledge

to help users understand its capabilities

When responding to a user’s inquiry, the app must allow users to

access relevant articles the AI model referenced to generate the answer

I also worked with the Product Manager to create the user flow of the app. This helped me understand which interfaces are needed for the application.

T H E U S E R F L O W

Users open the app to the home screen

Selects a product from the displayed product offerings

Enters a question related to the product into the chatbot

Clicks the send icon in the chatbot

Chatbot uses AI to generate a response and relevant articles

To ensure the user flow aligned with the Client Executive's vision, the Product Manager presented it for review. Once aligned, the Client Executive created a rough outline of the envisioned product interface.

D E S I G N E X E C U T I O N

After understanding Appian's design capabilities,

I converted the ideas into simple interfaces that focused on communicating the chatbot's capabilities and product offerings.

Below are some of the interfaces I created for the chatbot using Appian. As these designs will be presented as a product demonstration to one of our healthcare clients, my goal was to keep the design to-the-point and emphasize the product capabilities.

P R O D U C T C A T A L O G

Chat Interface

This is where the user will communicate and ask questions to the chatbot.

Product

Offerings

These cards indicate the topics users can inquire about.

Each card’s subtext outlines the specific data the model provides on the topic.

Although I aimed to keep the design simple, I incorporated visual elements to reduce text heaviness and support efficient decision-making. To do this, I:

Added icons to represent different products, enabling users to quickly scan the catalog and identify their desired product.
Adhered to the client’s branding by using the their design system to ensure visual consistency and organized relevant information into cards for clarity.

R E S U L T S

Results

Once the user requests for information, a results page will pop up with relevant articles used to generate the response.

As the team was still in the process of getting the sample article from the client, I was instructed to leave the text area blank.

S C O P E C H A N G E

After designing the interface for the Product Catalog and Results Page, I conducted a design review where I presented the interfaces to the team for feedback.

Even though the designs met the teams expectations of a simple and intuitive experience,

they realized that the products offered may be limited due to the constantly evolving medical industry.

To conquer this, the team brainstormed a solution that allows client medical experts to train the model themselves.

In turn, the team needed additional screens designed.

To ensure that the prompt tuning screens meet healthcare regulations and addresses all aspects of the new scope, I worked with the team to define the design requirements.

P R O M P T T U N I N G S C R E E N R E Q U I R E M E N T S

The model must be trained by the accurate healthcare professional to meet FDA and HIPAA regulations. Therefore, each task should be assigned to a particular user.

The user to which the task is assigned to, must be able to accept or reassign the task

(if the task is assigned to the incorrect user)

After, I converted these requirements into functional designs.

A C C E P T T A S K

R E A S S I G N T A S K

Reassign Task

If the task was assigned to the incorrect user, the user can reassign the task using the 'REASSIGN' button.

D E S I G N T O E N G I N E E R H A N D O F F

After I aligned the new Prompt Tuning interfaces with the Product Manager and Client Executive,

I worked with the engineer to implement the designs and encountered a few limitations with Appian.

A P P I A N L I M I T A T I O N S

Limited color options for container boxes and headers

Limited font sizes and styles

I re-visited the client's design system and Appian's design capabilities to find alternative approaches that provided a similar user experience.

D E S I G N S O L U T I O N S T O A P P I A N L I M I T A T I O N S

Limited color options for container boxes and headers

Used colors in Appian that align with the client's design system, while matching the color group of the original designs.

Limited font sizes and styles

Added spaces between chats, headers, and paragraphs to differentiate visual elements and types of content. Used different levels of bolding for titles, body text, and labelling.

T H E F I N A L P R O D U C T

After the engineer and I aligned on the design changes with the team, we released the beta version of the product and presented a demo to our client.

F I N A L R E S U L T S P A G E

Due to NDA purposes not all product interfaces are shown.

T H E R E S U L T

After presenting the demo, the client requested us to build the full product which resulted in a:

$ 2,000,000

deal with the client

and generated demand from

2 Fortune 500

Healthcare Clients

L E S S O N S L E A R N E D

While designing interfaces during the product discovery phase, I discovered a key insight about creating effective user experiences early in the product lifecycle:

a straightforward design that highlights product capabilities is often more effective than complex, visually elaborate interfaces.

The product discovery phase is meant to test ideas at a minimal cost, helping determine which products or features to build. By prioritizing functionality over high-fidelity visuals at this stage, we can conserve time and resources allowing us to make better-informed decisions and drive stronger business outcomes.