Challenges
Deploy Fine tuned LLM model trained on medical data to support clinicians workflows in a native mobile app for clinicians to efficiently improve quality of care.
*New workflow adoption, human and AI in the loop
*Seamless transitions from mobile to web (Emergency and primary care settings)
*LLM constraints based on relevant medical data, first mover into Emergency care
After integration with EHR, clinicians can access their pre-populated schedule and automate or manually make medical note updates based on personal preferences.
Defining success
Reach high adoption and usage with pilot groups to improve the mvp, showing that we understand the medical context, and positive feedback will allow scaling across multiple healthcare systems.
Defining success
In monitoring engagement, drop off occurred anytime there was a technical issue with mobile devices, clincians were used to using a desktops during patient visit, etc.
Strategy
Build a 1:1 from mobile app, with successful features and more editing capabilities because
most clinicians don't want to edit on mobile, but will make final edits on desktop due to current workflow with EHR.
Defining success
In monitoring engagement, drop off occurred anytime there was a technical issue with devices. Clinicians were used to desktop workflows during patient visits. So our team prioritized for a modular workflow allowing users to switch as needed between contexts.
Pivot
User research and feedback from clinicians, detailed that clinicians wanted an AI solution with human support in the loop, so product pivoted to a solution where an MDS could access, and update generated notes.
Defining success
Empowering clinicians to maximize AI, but still have support where needed through quick in app messaging
Prioritization
Design pushed to improve the UX based on feedback from Customer Success stakeholders and MDS needs. Allowing priorities and resources to shift towards maximizing NPS, our team achieved this by making onsite visit to various clinics.
Defining success
After account integration with EHR, clinician can access their pre-populated schedule and make updated as needed.
ML improvements
As LLMs improved we were able to fine tune the models to provide context specific recommendations to improve note quality with a few taps, and output summarized note formats.
Defining success
Empowering clinicians to maximize AI, but still have support where needed through quick in app messaging
Scaling
I communicated insights that clinicians found that the product was trying to constrain them to use it a certain way, once we added personalization features such as templates, and preferences and settings. We were able to let er rip, allowing clinicians to find their own workflow with minimal guardrails!
Defining success
Custom built CRM adaptable to each healthcare systems needs
Personalization
Augmedix was also going through a rebrand during the last quarter of the year, where I supported marketing and branding efforts in how we communicate the value of the product,
and even made space for a design system to keep it all consistent!
If you have questions those processes or this case study, feel free to reach out. ac.design.px@gmail.com
Defining success
MDS provided with real time updates based on clinicians updates in real time. Offering tools to increase support and efficiency
Notes
Augmedix was also going through a rebrand during the last quarter of the year, where I supported marketing and branding efforts in how we communicate the value of the product,
and even made space for a design system to keep it all consistent!
If you have questions those processes or this case study, feel free to reach out. ac.design.px@gmail.com
Next up
Contact
studio.acas
2025
Augmedix Web App
Role
Product / Design Systems Designer
Timeline
OCT '23 - NOV '25
Results
Results
Scaled to 5000+ clinicians, leading to 40 -
50% increase in AI note creation at HCA,
the nation's largest healthcare system