Healthcare
AI
Web Platform
How might we redesign the discharge summary workflow for the AI era?
Redesigning Foster Health's discharge summary workflow by integrating AI-assisted generation, editing, and sharing into an existing clinical experience, helping doctors reduce documentation time while maintaining complete control over the final summary.
Role
Product Designer
Company
Foster Health
Platform
WebApp
Status
Live
Year
2026
Background
Foster Health is an AI-powered clinical documentation platform built to streamline healthcare workflows. It helps clinicians capture consultations, generate structured clinical notes, and reduce administrative effort through AI-assisted documentation.
As the platform evolved, AI successfully supported individual consultation summaries. However, one critical stage of the clinical workflow still relied on manual effort: creating discharge summaries.
- Before the redesign · Existing AI summarization workflow
The problem
The existing AI workflow was built to summarize individual clinical transcriptions, making it useful for documenting a single consultation.
However, clinicians preparing discharge summaries often needed to review multiple consultations, combine information from different notes, and manually create the final document. While AI reduced effort at the note level, the overall discharge documentation process remained manual and fragmented.
AI helped summarize individual consultations, but preparing a discharge summary still meant manually reviewing multiple notes and bringing everything together.
- Insight from stakeholder discussions
As consultations accumulated throughout a patient's journey, documentation became increasingly disconnected.
The absence of a dedicated discharge summary workflow meant clinicians had to leave the AI-assisted experience and manually assemble information before sharing it with patients.
Key findings
How might we support documentation beyond a single consultation?
Build a multi-consultation workflow
The existing AI summarized one consultation at a time, but clinicians needed to combine information across multiple encounters to create a complete discharge summary.
How might we reduce manual review and documentation effort?
Keep clinicians in control while reducing effort
Preparing summaries required reviewing multiple notes and manually consolidating information. AI should assist the process without removing clinical oversight.
How might we create one continuous documentation experience?
Connect the entire discharge workflow
Generating, reviewing, editing, and sharing discharge summaries should happen within one connected workflow instead of across disconnected screens.
Methodology
This redesign followed an iterative design process, combining research, exploration, validation, and continuous collaboration to transform an existing AI summarization experience into a complete discharge summary workflow.
Research
To understand the existing workflow, I analyzed the current product, observed how clinicians documented patient encounters, and collaborated with stakeholders to identify opportunities for improving discharge documentation.
Product Analysis
Gap Analysis
UI Audit
Product Review
User Research
Documentation Flow
Pain Points
Stakeholder Interviews
Design Research
Competitor Analysis
Information Architecture
Wireframes
Research Insights
The research uncovered four recurring insights that became the foundation for the redesigned discharge summary workflow.
Product Pain Points
Fragmented workflow
Scattered notes
High effort
Repeated reading
Poor navigation
User Needs
Quick overview
Fast review
Familiar layout
Clear hierarchy
Confidence in AI
Product Opportunities
Unified workflow
Multi-note selection
Context linking
Search & filtering
Summary preview
Stakeholder Priorities
AI-assisted workflow
Editable output
Faster documentation
Easy implementation
Scalable design
Workflow & Exploration
- Before the redesign · Existing AI summarization workflow
The existing workflow was optimized for individual consultations, but couldn't support end-to-end discharge documentation.
Redesigning Navigation
Explored multiple navigation patterns to improve note discovery, organization, and scalability as the product evolved.
However, Navigation was only one part of the problem. The underlying information structure also needed to evolve.
Restructuring Information Architecture
Navigation issues were symptoms of a deeper structural problem. Reorganizing how notes, folders, and workspaces were connected became the foundation of the redesign.
A scalable navigation system begins with a scalable information architecture.
Defining Note Interactions
Once the navigation structure was established, interaction patterns were redesigned to simplify note management and reduce unnecessary user actions.
Flexible note interactions reduced friction and supported a more efficient documentation workflow.
Redesigned Navigation
The redesigned navigation introduced hierarchy, folders, and scalable organization while keeping note discovery simple and predictable.
Structured navigation made it easier to manage growing consultation histories without increasing complexity.
Building the Foundation for AI-Assisted Documentation
With the platform foundation redesigned, Foster Health could now support more advanced clinical workflows. This created the foundation for an AI-assisted discharge summary experience.
With the workspace established, the next challenge was enabling clinicians to generate discharge summaries from multiple consultations instead of a single clinical note.
Designing the AI-assisted discharge workflow
With the documentation foundation in place, the next challenge was bringing information from multiple consultations into a single discharge summary, while keeping clinicians in control of the final output.
Step 1
Select notes
Step 2
Generate
Step 3
AI processing
Step 4
Review & edit
Step 5
Export
Selecting the right clinical context
A discharge summary can span multiple consultations, so clinicians needed a simple way to choose which notes should inform the AI-generated summary.
Clinicians stay in control of which clinical information is used before generation begins.
From selected notes to a discharge summary
Once the relevant consultations were selected, clinicians could generate a discharge summary using the combined context. The system processed the selected notes, generated a structured summary, and presented it for clinician review and editing before it could be exported or shared.
Turning clinical notes into a structured summary
The selected consultations were synthesized into a structured discharge summary, giving clinicians a single place to review the patient's clinical journey before finalizing the document.
AI handles the initial synthesis. The clinician remains responsible for the final document.
Solving the surrounding product problems
Export & Share
Workflow
Making the final summary easier to take forward
Before, the discharge workflow ended with the summary inside the workspace, leaving the final step of exporting or sharing disconnected from the process.
Alerts & Notifications
Feedback
Making critical system states visible
Important actions and system states could easily get lost during a clinical workflow. The new feedback patterns made these moments clearer without interrupting the task.
Support
Help
Getting help without leaving the workflow
Support was separated from the work clinicians were doing. Bringing it into the workspace meant help could be accessed without breaking the documentation flow.
Mobile adaptation
App
Bringing the workflow beyond the desktop
The core experience was built around a desktop workspace. The mobile adaptation focused on making essential patient information and actions accessible on smaller screens.
What changed
The redesign shifted discharge documentation from a fragmented, note-by-note process into a connected workflow built around the patient's clinical journey.
Before
AI was focused on individual consultation summaries
Clinicians had to manually bring information from multiple notes together
Patient history became difficult to organize as consultations accumulated
The discharge workflow was disconnected from the existing documentation experience
After
Multiple clinical notes can inform a single discharge summary
Patient information is organized through a clearer hierarchy
AI generates a structured starting point for the summary
Clinicians remain in control of the final output
Completed summaries can be exported or shared
What I learned
The hardest part wasn't designing the AI. It was designing everything around it.
Working on Foster Health made me realize that introducing AI into an existing product isn't just about generating a better output. The surrounding workflow, information architecture, interaction patterns, and user control all need to support that intelligence.
The project also changed how I approach complex product problems. Instead of immediately designing the new feature, I had to first understand why the existing system was breaking down, rebuild the foundation, and then introduce AI where it could actually reduce the work.
Let’s talk
over a cup of chai.
Hakesh Sundramoorthy