AI-Powered Healthcare Knowledge Assistant - VIA

0.1 Project Overview

 

Designing an AI-powered knowledge platform for healthcare professionals.

The project is an enterprise AI platform designed to help doctors, healthcare practitioners, and hospital professionals access organizational knowledge through conversational AI.

Healthcare organizations often maintain large amounts of information across multiple documents, research papers, medical reports, and internal data sources. Finding relevant information across these sources can be time-consuming.

The platform brings these knowledge resources together into a centralized AI-powered experience, allowing users to ask questions and receive contextual responses based on selected knowledge profiles.

The product consists of two interconnected web applications:

  1. Admin Portal: Used to create and manage AI profiles, configure AI models, and connect knowledge sources.

  2. AI Chat Application: Used by healthcare professionals to interact with AI, upload documents, select models, and manage conversations.

0.2 The Problem

 

Healthcare organizations manage information across multiple systems and document repositories.

Doctors and healthcare practitioners may need to refer to research papers, medical documentation, internal reports, and organizational knowledge to find specific information.

However, accessing and understanding information from multiple sources can create several challenges.

Key problems identified:

  • Information is distributed across multiple data sources.

  • Searching through large documents takes time.

  • Users need a centralized way to interact with organizational knowledge.

  • Different teams may require different knowledge contexts.

  • Standard AI chat interfaces do not necessarily understand an organization’s internal knowledge.

  • Sharing customized knowledge configurations between colleagues can be difficult.

Design opportunity: Create a centralized AI experience that allows users to access relevant organizational knowledge through customized profiles and conversational interactions.

0.3 Project Goals

 

The primary goal was to create an intuitive enterprise AI platform that simplifies access to organizational knowledge.

Goal

UX objective

Centralized knowledge

Bring documents and data sources into a unified experience

Customized AI

Allow administrators to create purpose-specific profiles

Easy interaction

Provide a familiar conversational interface

Model flexibility

Allow users to choose available AI models

Better accessibility

Make previous conversations easy to retrieve

Collaboration

Enable users to share knowledge profiles

Administration

Simplify AI profile and model management

0.4 Understanding the Product Architecture

 

One of the important aspects of this project is its dual-application architecture.

Both applications serve different user groups but work together as one ecosystem.

Application 1 — Admin Portal

The administrative application provides tools to configure and manage the AI ecosystem.

Key modules:

  • AI profile creation and management.

  • Knowledge document management.

  • Enterprise data source selection.

  • AI model configuration.

  • Default model assignment.

  • Starter prompt configuration.

  • Profile sharing and management.

Application 2 — AI Chat Platform

The user-facing application enables healthcare professionals to interact with configured AI profiles.

Key modules:

  • User authentication.

  • AI profile selection.

  • AI model selection.

  • Conversational AI interface.

  • Multiple document uploads.

  • Previous conversation history.

  • Shared AI profiles.

05. User Roles

 

The platform supports different user responsibilities.

User type

Responsibilities

Administrator

Create profiles, manage knowledge sources, configure models

Healthcare professional

Select profiles, interact with AI, upload documents

Colleague / Shared user

Access profiles shared by other users, subject to permissions

06. Core UX Features and Design Solutions

 

The platform supports different user responsibilities.

 
6.1 AI Profile Management
 

Challenge: Different departments and healthcare professionals need different knowledge contexts.

Design solution:

The platform provides a customizable AI profile management system.

Administrators can create multiple profiles based on specific purposes, departments, or knowledge requirements.

Each profile contains:

  • Profile name.

  • Profile description.

  • Predefined starter prompts.

  • Uploaded knowledge documents.

  • Selected enterprise data sources.

  • Default AI model.

This approach allows administrators to organize knowledge into separate, purpose-specific AI experiences.

 

6.2 Knowledge Source Management
 

Challenge: Organizational knowledge exists across multiple documents and internal data sources.

Design solution:

The profile creation workflow allows administrators to combine two types of knowledge resources:

A. Document-based knowledge

  • Upload individual documents.

  • Upload multiple documents.

  • Add medical reports.

  • Add research papers and other relevant files.

B. Enterprise data sources

  • Select existing company data sources.

  • Associate relevant sources with individual profiles.

  • Configure profiles using existing organizational knowledge.

This gives administrators flexibility to create AI profiles using both uploaded documents and existing enterprise information.

 

6.3 AI Model Configuration
 

Challenge: Different AI models may be available within the organization, and users need flexibility when interacting with AI.

Design solution:

The platform supports AI model selection at two levels:

  • Admin Portal: Administrators configure available models and assign a default model to an AI profile.

  • AI Chat Application: Users can select an available AI model directly from the chat interface.

This creates flexibility while maintaining centralized model configuration.

 

6.4 Conversational AI Interface
 

Challenge: Healthcare professionals need a familiar and straightforward way to interact with organizational knowledge.

Design solution:

A conversational interface allows users to communicate with the AI using natural language.

Users can:

  • Ask questions.

  • Receive AI-generated responses.

  • Continue conversations.

  • Upload supporting documents.

  • Interact using selected AI profiles.

  • Choose available AI models.

The conversational approach reduces the need for users to manually navigate multiple knowledge repositories.

 

6.5 Multiple Document Uploads
 

Challenge: Users may need to provide additional documents during a conversation.

Design solution:

The chat interface supports document uploads, including multiple files, so users can provide additional information while interacting with AI.

This complements the documents already associated with the selected AI profile.

 

6.6 Chat History and Session Management
 

Challenge: Users may need to revisit previous conversations instead of starting from scratch every time.

Design solution:

The platform stores conversation sessions in chat history.

Users can:

  • View previous sessions.

  • Reopen conversations.

  • Continue previous discussions.

  • Organize their interaction history through the conversation list.

 
6.7 Profile Sharing and Collaboration
 

Challenge: Customized AI knowledge configurations may be useful to multiple colleagues.

Design solution:

Users can share AI profiles with colleagues, allowing others to access and use the same configured knowledge experience, subject to the platform’s access permissions.

This supports knowledge reuse and collaboration within healthcare organizations.

 

07. End-to-End User Journey
 

The following journey explains how the two applications work together.

Administrator workflow

  1. Log in to the Admin Portal.
  2. Create a new AI profile.
  3. Enter the profile name and description.
  4. Configure predefined starter prompts.
  5. Upload documents and select enterprise data sources.
  6. Select the default AI model.
  7. Save and make the profile available to authorized users.

Healthcare professional workflow

  1. Log in to the AI Chat Application.
  2. Select an available AI profile.
  3. Use the default AI model or choose another available model.
  4. Start a conversation using natural language.
  5. Upload additional documents if required.
  6. Receive AI-generated contextual responses.
  7. Access previous conversations from chat history.
  8. Share useful AI profiles with colleagues.
     
08. UX Design Considerations
 

The product combines administration, AI interactions, document management, and enterprise knowledge configuration. Therefore, the design needs to address several UX considerations.

UX area

Design consideration

Information architecture

Separate admin and end-user functionality

Navigation

Make profiles, documents, and chat history easy to access

Consistency

Maintain consistent interaction patterns

Feedback

Provide clear upload and processing status

Model selection

Make the currently selected AI model visible

Profile selection

Clearly communicate the selected knowledge context

Collaboration

Make sharing and access permissions understandable

Error handling

Provide understandable error and recovery messages

 
09. My Role and Responsibilities
 

As a UI/UX Designer, my contribution to this project focused on designing an intuitive user experience for a complex enterprise AI platform.

My responsibilities included:

  • Understanding the product requirements and business objectives.

  • Structuring the user experience across two web applications.

  • Designing the information architecture and navigation.

  • Creating user flows for administrators and healthcare professionals.

  • Designing AI profile creation and management workflows.

  • Designing document upload and data source selection experiences.

  • Designing AI model selection interactions.

  • Designing the conversational AI interface.

  • Designing chat history and profile-sharing experiences.

  • Creating consistent UI patterns across the platform.

Only retain responsibilities that you personally performed.

 
10. Design Challenges
 

Some of the key UX challenges associated with this product were:

Challenge 1: Managing two interconnected applications

The admin portal and chat application have different purposes and user workflows.

UX approach: Separate the administrative configuration experience from the conversational end-user experience while maintaining consistency across the product ecosystem.

Challenge 2: Simplifying complex profile creation

Creating an AI profile requires multiple configurations, including prompts, documents, data sources, and AI models.

UX approach: Organize these settings into understandable sections within the profile creation workflow.

Challenge 3: Making AI interactions accessible

Users may not be familiar with different AI models or knowledge configurations.

UX approach: Provide a familiar chat experience with visible profile and model selection.

Challenge 4: Combining multiple knowledge sources

Users need access to information from uploaded documents and existing company data sources.

UX approach: Provide a unified profile configuration experience for managing both types of knowledge resources.

 

11. Project Impact
 

Since you haven’t provided actual usage statistics or business metrics, I recommend presenting the expected product value rather than inventing numerical results.

The platform is designed to:

  • Reduce friction when accessing organizational knowledge.

  • Simplify interaction with multiple knowledge sources.

  • Provide reusable, customized AI knowledge profiles.

  • Support flexible AI model selection.

  • Improve knowledge accessibility across teams.

  • Enable collaboration through profile sharing.

  • Provide a centralized conversational AI experience.

 
12. Key Learnings
 

Working on this type of enterprise AI product provides an opportunity to understand the UX requirements of AI-powered applications beyond conventional chatbot interfaces.

Key design learnings include:

  • Designing for AI requires more than creating a chat interface.

  • Knowledge architecture plays an important role in AI product usability.

  • Complex administrative workflows need clear information architecture.

  • AI model selection should be understandable and accessible.

  • Enterprise products need to consider collaboration and permissions.

  • Document and data source management are important parts of the overall AI experience.

Thank You