Enhancing Quality,
Reducing Waste & Costs
Enhancing Quality,
Reducing Waste & Costs
Enhancing Quality,
Reducing Waste & Costs
Enhancing Quality,
Reducing Waste & Costs
Up to 99% precision in pharmaceutical visual inspection, lowering costs, reducing waste, improving efficiency & patient safety.
Up to 99% precision in pharmaceutical visual inspection, lowering costs, reducing waste, improving efficiency & patient safety.
Up to 99% precision in pharmaceutical visual inspection, lowering costs, reducing waste, improving efficiency & patient safety.
Up to 99% precision in pharmaceutical visual inspection, lowering costs, reducing waste, improving efficiency & patient safety.
99% Defect detection accuracy
The AI system identifies nearly all defects, outperforming both manual/ machine inspections
75% Faster task completion
From 4 min to under 1 min per image labeling task after UI an AI improvements
100% User satisfaction
Visual inspectors reported increased satisfaction with the new interface
Context
1% error rate can cost €50M. Automated inspection machines in pharma often misidentify defects, leading to wasted resources (false positives) and safety risks (false negatives). Human visual double-checks are unreliable, increasing costs and compliance risks. To address these issues, Körber pharma applies AI to automate visual inspection, accurately detecting defects in vials and ampules.
1% error rate can cost €50M. Automated inspection machines in pharma often misidentify defects, leading to wasted resources (false positives) and safety risks (false negatives). Human visual double-checks are unreliable, increasing costs and compliance risks. To address these issues, Körber pharma applies AI to automate visual inspection, accurately detecting defects in vials and ampules.
Project Overview
As the UI/UX designer, I led the end-to-end design of a SaaS platform for pharmaceutical quality control, collaborating with stakeholders, engineers, data scientists, product owners, project managers, and users. This case study focuses on the Labeling app—a key part of this ecosystem—designed to improve labeling accuracy, efficiency, and user experience in a high-compliance environment.
As the UI/UX designer, I led the end-to-end design of a SaaS platform for pharmaceutical quality control, collaborating with stakeholders, engineers, data scientists, product owners, project managers, and users. This case study focuses on the Labeling app—a key part of this ecosystem—designed to improve labeling accuracy, efficiency, and user experience in a high-compliance environment.
1
1
1
1
Discovery/Research
Discovery/Research
Discovery/Research
To understand the real challenges in pharmaceutical visual inspection, I conducted qualitative research with 6 end users, combining user interviews, workflow analysis, and stakeholder workshops. Using affinity mapping, I grouped insights into four key themes: Human Factors & Cognitive Load, Safety & Compliance, Efficiency, and Sustainability & Waste Reduction. These findings revealed user pain points and business needs, and helped align expectations across teams. Based on these insights, I created a persona to guide design decisions.
To understand the real challenges in pharmaceutical visual inspection, I conducted qualitative research with 6 end users, combining user interviews, workflow analysis, and stakeholder workshops. Using affinity mapping, I grouped insights into four key themes: Human Factors & Cognitive Load, Safety & Compliance, Efficiency, and Sustainability & Waste Reduction. These findings revealed user pain points and business needs, and helped align expectations across teams. Based on these insights, I created a persona to guide design decisions.
To understand the real challenges in pharmaceutical visual inspection, I conducted qualitative research with 6 end users, combining user interviews, workflow analysis, and stakeholder workshops. Using affinity mapping, I grouped insights into four key themes: Human Factors & Cognitive Load, Safety & Compliance, Efficiency, and Sustainability & Waste Reduction. These findings revealed user pain points and business needs, and helped align expectations across teams. Based on these insights, I created a persona to guide design decisions.
Key Insights
Key Insights
Key Insights
User pain points
❌ High cognitive load and fatigue from repetitive labeling tasks
❌ Difficulty distinguishing subtle defects
❌ Frustration with lack of feedback
Business impact
❌ Increased operational costs due to high false reject rates
❌ Delays in production and quality assurance
❌ Compliance risks and potential threats to patient safety
2
2
2
2
Define
Define
Define
After gathering insights from users and stakeholders during the research phase, I synthesized the findings to clearly define the core problem and project objectives. This stage focused on translating user pain points and business needs into actionable goals, setting a clear direction for the design process.
After gathering insights from users and stakeholders during the research phase, I synthesized the findings to clearly define the core problem and project objectives. This stage focused on translating user pain points and business needs into actionable goals, setting a clear direction for the design process.
After gathering insights from users and stakeholders during the research phase, I synthesized the findings to clearly define the core problem and project objectives. This stage focused on translating user pain points and business needs into actionable goals, setting a clear direction for the design process.
Challenges
Repetitive tasks and fatigue lead to errors
AI struggles with subtle defect classification
Onboarding new operators is slow and costly
Repetitive tasks and fatigue lead to errors
AI struggles with subtle defect classification
Onboarding new operators is slow and costly
Problem Statement
Problem Statement
Problem Statement
Manual visual inspection in pharmaceutical quality control is prone to human error, fatigue, and inconsistency—leading to unreliable defect detection, increased costs, and potential safety risks.
Manual visual inspection in pharmaceutical quality control is prone to human error, fatigue, and inconsistency—leading to unreliable defect detection, increased costs, and potential safety risks.
Manual visual inspection in pharmaceutical quality control is prone to human error, fatigue, and inconsistency—leading to unreliable defect detection, increased costs, and potential safety risks.
Project Goal
Project Goal
Project Goal
Accuracy
Accuracy
Accuracy
Minimize human error in defect labeling
Minimize human error in defect labeling
Minimize human error in defect labeling
Safety
Safety
Safety
Improve reliability and consistency of inspections
Improve reliability and consistency of inspections
Improve reliability and consistency of inspections
Simplification
Simplification
Simplification
Reduce user fatigue during inspection tasks
Reduce user fatigue during inspection tasks
Reduce user fatigue during inspection tasks
Waste Reduction
Waste Reduction
Waste Reduction
Support sustainability by reducing waste and operational costs
Support sustainability by reducing waste and operational costs
Support sustainability by reducing waste and operational costs
Solution definition
Solution definition
Solution definition
Devices
For detailed tasks like defect labeling (zoom, drag, bounding boxes), screen size is critical for usability and accuracy.
For detailed tasks like defect labeling (zoom, drag, bounding boxes), screen size is critical for usability and accuracy.
For detailed tasks like defect labeling (zoom, drag, bounding boxes), screen size is critical for usability and accuracy.
Monitors / Laptops
Designed for efficient multitasking and labeling, with ergonomic setups to reduce strain. Recommended for stationary work environments. Minimum screen size: 13 inches.
Tablet
Lightweight and portable, ideal for mobile or travel use. Requires a stylus or pen for precise input. Minimum screen size: 10.5 inches.
Proposed Features
Enhanced Labeling Tools
Enhanced Labeling Tools
Enhanced Labeling Tools
Track progress
Zoom and drag
Bounding box
Consistency Checks
Consistency Checks
Consistency Checks
Precision scores report
Audit trail log
Precision scores report
Audit trail log
AI-Assisted Labeling
AI-Assisted Labeling
AI-Assisted Labeling
AI suggestions
Editable by user
3
3
3
3
Ideation & Solution Exploration
Ideation & Solution Exploration
Ideation & Solution Exploration
During the ideation phase, the Product Owner was closely involved in shaping and validating our assumptions. Together, we defined key hypotheses to guide design decisions and prioritize features for testing.
During the ideation phase, the Product Owner was closely involved in shaping and validating our assumptions. Together, we defined key hypotheses to guide design decisions and prioritize features for testing.
During the ideation phase, the Product Owner was closely involved in shaping and validating our assumptions. Together, we defined key hypotheses to guide design decisions and prioritize features for testing.
Hypotheses
Hypotheses
Hypotheses
If we provide real-time feedback on labeling progress, users will feel more motivated and engaged
Streamlining the labeling workflow will reduce cognitive load and improve overall task efficiency
Offering a comprehensive defect library will help users make more consistent labeling decisions
First Drafts
First Drafts
First Drafts
4
4
4
4
Validation & Refinement
Validation & Refinement
Validation & Refinement
I led refinement workshops with the full team to gather input and address technical challenges. This collaborative approach ensured the design was feasible and aligned with both user needs and technical constraints. To validate usability and identify pain points, users tested key features through real-world tasks using a mid-fidelity prototype. At least 3 iterations were conducted based on feedback, refining the design and preventing costly issues later.
I led refinement workshops with the full team to gather input and address technical challenges. This collaborative approach ensured the design was feasible and aligned with both user needs and technical constraints. To validate usability and identify pain points, users tested key features through real-world tasks using a mid-fidelity prototype. At least 3 iterations were conducted based on feedback, refining the design and preventing costly issues later.
I led refinement workshops with the full team to gather input and address technical challenges. This collaborative approach ensured the design was feasible and aligned with both user needs and technical constraints. To validate usability and identify pain points, users tested key features through real-world tasks using a mid-fidelity prototype. At least 3 iterations were conducted based on feedback, refining the design and preventing costly issues later.
Iterations feedback
Iterations feedback
Iterations feedback
User pain points
❌ 1. Confusion between defects and selections
❌ 2. Marking "No Defect" per box was repetitive
❌ 3. No way to revise previous labels
Technical constraints
➔ 1. None
➔ 2. None
➔ 3. Limited to editing the last 3 images
Design solution
➔ 1. Applied distinct colors and overlays for clarity
➔ 2. Added a general "No Defect" button on footer
➔ 3. Added options to revisit and edit previous labels
1
1
2
2
3
3
Final Design*
Final Design*
Final Design*
Design Principals Applied
Visibility & Feedback
Visibility & Feedback
Visibility & Feedback
Task progress indicator
Clear status indicators for defect detection
Task progress indicator
Clear status indicators for defect detection
Task progress indicator
Clear status indicators for defect detection
Accessibility & Equity
Accessibility & Equity
Accessibility & Equity
Efficiency & Workflow
Efficiency & Workflow
Efficiency & Workflow
Use shortcuts
Predictive inputs to improve speed.
Use shortcuts
Predictive inputs to improve speed.
Use shortcuts
Predictive inputs to improve speed.
*High-Fidelity Prototype (Labelling app)- Due to compliance and data protection policies, these visuals and data have been modified for showcasing purposes.
*High-Fidelity Prototype (Labelling app)- Due to compliance and data protection policies, these visuals and data have been modified for showcasing purposes.
5
5
5
5
Implementation
In this phase, I focused on documenting design specifications for developers and prioritizing features based on user needs, business goals, and technical feasibility. This ensured a clear handoff and alignment across the team for a smooth development process.
KPI's
KPI's
KPI's
99% Defect detection accuracy
99% Defect detection accuracy
99% Defect detection accuracy
The AI system now identifies nearly all defects, outperforming both manual inspections and previous machine systems.
The AI system now identifies nearly all defects, outperforming both manual inspections and previous machine systems.
The AI system now identifies nearly all defects, outperforming both manual inspections and previous machine systems.
75% Faster task completion
75% Faster task completion
75% Faster task completion
From 4 min to under 1 min per image labeling task after UI an AI improvements
From 4 min to under 1 min per image labeling task after UI an AI improvements
From 4 min to under 1 min per image labeling task after UI an AI improvements
100% User satisfaction
100% User satisfaction
100% User satisfaction
6/6 of labelers reported increased satisfaction with the new interface
6/6 of labelers reported increased satisfaction with the new interface
6/6 of labelers reported increased satisfaction with the new interface
key lessons
key lessons
key lessons
Reflection
Small usability improvements matter: Even minor interface enhancements can significantly boost accuracy, efficiency, and user satisfaction.
Iterative testing drives better outcomes: Regular feedback and multiple design iterations help uncover pain points early and refine the product.
Stakeholder alignment accelerates success: Ongoing communication with stakeholders ensures the solution meets both business and regulatory requirements.
Small usability improvements matter: Even minor interface enhancements can significantly boost accuracy, efficiency, and user satisfaction.
Iterative testing drives better outcomes: Regular feedback and multiple design iterations help uncover pain points early and refine the product.
Stakeholder alignment accelerates success: Ongoing communication with stakeholders ensures the solution meets both business and regulatory requirements.
Next steps
Leverage these learnings to design scalable, inclusive solutions for other high-compliance industries while continuing to prioritize sustainability and usability.
Leverage these learnings to design scalable, inclusive solutions for other high-compliance industries while continuing to prioritize sustainability and usability.
Leverage these learnings to design scalable, inclusive solutions for other high-compliance industries while continuing to prioritize sustainability and usability.
Portfolio
Other projects
Other projects
Other projects
Bite - Frozen meal
E-commerce ∙ Case study
Photobooth redesign
Photobooth redesign
Photobooth redesign
Software ∙ Case Study
Software ∙ Case Study
Software ∙ Case Study
AOP Punk Rock
E-commerce ∙ Events ∙ UI
See all projects
