Copyright © 2025 All rights reserved. Built with love in Texas.
Branding Product SaaS Web App
About a Project
This internal tool was built to support data scientists and machine learning engineers working on custom clinical AI models. It enables fast experimentation, model fine-tuning, and comparison by combining ambient transcription inputs with chat-based interactions and structured clinical note generation.
The platform brings together transcript uploads, AI outputs, prompt testing, and document generation in one streamlined interface.
The Challenge
Design a system that would allow internal AI teams to:
Upload large transcription datasets for model input
Interact with models using natural language prompts and medical context
View model outputs as structured notes (SOAP, clinical summaries, referrals)
Compare output across model versions and prompt variations
Enable rapid iteration and model performance tuning in real-world clinical formats
Key Features
Transcript Upload & Context Preservation – Designed for 2MB+ files, preserving medical context throughout prompt-response iterations
Chat Interface for Prompt Engineering – Allows iterative dialogue with fine-tuned LLMs for testing output reliability
Note-Type Toggle – One-click switching between SOAP, Clinical Note, and Referral Note output formats
Side-by-Side Compare Mode – Built for validating changes between prompt versions or model iterations
🖥️ UX & UI Highlights
Left Panel: Conversation index for organized model runs
Center Workspace: Chat + upload stream showing live input/output flow
Right Panel: Real-time document viewer with generated transcription and notes
Dark Theme UI: Purpose-built for extended sessions, reducing visual fatigue
Clear iconography and progressive disclosure for low-friction navigation
🔬 Product & UX Outcomes
Empowered internal teams to test AI model variants faster
Created a shared space for prompt engineers and clinical SMEs to validate results
Structured MVP with clear epics and user stories focused on iteration speed and usability
Reduced time-to-feedback on model fine-tuning from days to minutes
🧪 Clinical-AI Alignment
The design bridges technical workflows with clinical usability by presenting AI output in familiar formats—helping internal teams validate outputs as real, chart-ready notes.