Tandem Diabetes Care — t:slim insulin pump

A digital insulin pump with a black display showing battery level, time, date, insulin levels, and options menu

Executive summary

As the first product designer hired at Tandem, I worked directly with company founders and executives to define a user experience (UX) and design strategy in a deeply technical, FDA-regulated space.

  • I built the product design function from the ground up and led UX strategy, design, and research for the t:slim Insulin Pump, the first touchscreen insulin pump on the market.

  • This FDA-cleared, award-winning medical device redefined design and usability in the diabetes tech industry and became the company’s flagship product.

  • My leadership was instrumental in driving product success, market differentiation, and company growth from early-stage startup to IPO.

Outcomes

  • Winner of two Medical Device Excellence Awards:

    There was much praise in the Diabetes and medical community for the positive user experience and innovative functionality of the t:slim insulin pump. Awards received:

    The Bronze award

    Reader’s Choice award

  • 27% reduction in time to train patients

    Research has shown a 27% reduction in time to train patients on the t:slim versus the Medtronic MiniMed Paradigm Revel pump

  • 65% reduction in use errors vs the competition

    Research has shown a 65% reduction in use errors on the t:slim versus the Medtronic MiniMed Paradigm Revel pump

  • Innovative insulin on board indicator

    Insulin on board indicator provides important patient blood glucose information about remaining carbohydrates on board, not previously available in a consumer pump

  • Sophisticated error prevention

    Detailed and sophisticated error handling and active alerts increased patient confidence and engagement with insulin therapy

  • Convenient personal profiles for various settings

    Personal profiles allows a covenient way to preset different lifestyle scenarios, allowing the pump to be adjuted to your needs on the fly

My role and leadership impact

✅ Hired and led the first product design team at Tandem

✅ Created design frameworks and interactions that became the foundation of the product experience

✅ Provided leadership across cross-functional teams including product, engineering, clinical, and regulatory

✅ Championed design strategy with executive leadership, FDA, and medical professionals contributing to IPO readiness

✅ Defined and executed the design for every aspect of the t:slim product experience

Business & Strategic Objectives

  • Launch the industry’s first consumer-grade insulin pump interface

  • Differentiate in a market dominated by legacy medical devices through superior design and usability

  • Ensure full compliance with FDA human factors (HE75) and safety guidelines

  • Establish a scalable UX function and design culture within a highly regulated, engineering-led environment

White puzzle pieces scattered on a white surface.

Design

Navigating the t:slim

The primary functionality of the t:slim is delivering a bolus of insulin to the body to cover carbohydrate intake. As such, this is the primary action on the home screen.

The Options menu provides additional functionality that is useful and convenient, but not primary to the core experience.

Delivering a Bolus

A bolus is an injection of insulin to the body. The Bolus features in the t:slim allowed users to enter the amount of carbohydrates they plan to eat, and the pump would automatically calculate the appropriate dose based on your insulin to carbohydrate ratio.

If the patient adds a blood glucose value, the insulin dose accounts for a high or low reading, which could restrict the amount included in the dose as a safety precaution.

Personal profiles

The personal profiles features enabled very quick training and onboarding for both nurse educators and patients.

Key Contributions & Outcomes

🧠 Human-Centered Strategy in a Regulated Industry

  • Conducted deep stakeholder discovery with patients, nurses, endocrinologists, and execs

  • Balanced human needs, safety protocols, and technical constraints in every design decision

  • Created a custom interaction framework for a highly constrained mobil infusion pump

🔄 Groundbreaking Design in Constrained Hardware

  • Designed a touchscreen insulin delivery workflow that prioritized safety and clarity

  • Developed reusable UI patterns that reduced development overhead and increased consistency

  • Created Personal Profiles, a feature that reduced setup friction and improved patient outcomes

  • Quick bolus function provided streamlined insulin delivery without the UI

  • Extended bolus feature allows bolus to be incrementally delivered over time.

  • Insulin on Board indicator provided critical information about how insulin is being used by the patient.

🔬 Research-Driven FDA Compliance

  • Led formative research, personas, and ethnographic studies

  • Collaborated with human factors and regulatory teams to meet FDA submission standards

  • Ran extensive usability testing with Type 1 Diabetes patients and clinical educators

  • Delivered a system that passed summative testing with minimal revisions

💼 Enterprise Impact

  • t:slim became Tandem’s flagship product, directly contributing to company growth and IPO

  • UX foundation I established is still in use across current Tandem product lines

  • Helped Tandem compete with and surpass legacy players in usability and innovation perception

  • Product received multiple industry awards, becoming a new standard in diabetes tech

Want to go deeper?

Keep scrolling for more detail about this case study.

If you are ready to talk, send me a message.

Discovery insights

  • Look and feel

    In partnering with Market Research and Sales, we discovered the look & feel of the device is very important to users of insulin pumps. Patients’ are already self-conscious about their disease, so they do not want to call more attention to themselves.

  • Awareness

    People don’t understand all of the inputs for calculating their own insulin delivery.

  • How insulin is metabolized affects potential outcomes

    People don't understand insulin on board and how it is metabolized because current pumps do not give them good enough feedback or education during use.

  • Existing solutions are difficult to use

    Navigation is cumbersome and information is not organized in an intuitive way.

  • Intimidating language

    Alerts, alarms tend to be scary and intimidating, without help for resolving the issue.  

User research

  • Research overview

    I interviewed three type-1 diabetes patients, two nurse educators, and one Endocrinologist. I also leveraged focus group and market research data being conducted by the market research team. We developed a set of personas based on this research.

    The audience for an FDA regulated infusion pump is much wider than a typical consumer product.  It is designed specifically for patients of type-1 diabetes, but because this device was a prescribed by a physician and will be trained by a nurse educator, they must be included as secondary users to ensure the right amount of information and functionality, as well as their involvement in assessing a patient’s pump data.  The range of demographics is also fairly wide since the disease does not discriminate by gender, race or personality type.

  • User goals

    Confidently use and trust a pump is accurate and safe

    Easier way to understand how insulin degrades in the body once consumed

    Easy way to give a fast or temporary bolus

    Integration with continuous glucose monitoring (CGM) systems

    Device that looks more consumer than medical

    Avoid needles

  • User challenges

    Finger dexterity may be compromised in some patients

    Vision may be compromised in some patients

    UI needs to be accessible to color blind patients

    In hypoglycemic states, users may be in an emergency and need help

  • Next steps

    I developed personas from this and other interviews with nurses, physicians, doctors, sales, and educators.

Competitive analysis

Next I did research on other competing products to establish the baseline set of requirements for a product in this space. I partnered with market research, did online research, ordered several products and evaluated their out of box experience and overall first impressions. I also conducted interviews with patients who either own competing pumps or worked for competing companies to evaluate the primary audience's delighters and pain points with the different products.

Task analysis

I conducted task analyses on the core task flows for all of the competing products to evaluate the baseline against which similar feature may be measured for equivalency.

This included creating walk throughs or flow diagrams identifying the primary tasks and decisions users had to make to accomplish the most common tasks.

Use cases

Once the user personas / audience were identified, I worked with each audience to identify the primary use cases for each audience.

Leveraging the competitive analysis and the audience analysis, I produced documentation of the use cases identifying the step by step process for each scenario.

This effort was valuable for both the design of the user interface and documenting of test cases for the software development team.

Information Architecture

Based on the research, I began sketching out a hierarchy of features based on user and nurse educator feedback.  The competitive analysis gave me insight into how existing systems are structured, I took that as a starting point and began grouping information based on the desired patient workflow that I learned from interviewing medical professionals. 

After several card sorting exercises with 4-5 users, we arrived at an IA that strengthens the relationship between the pump and insulin delivery settings.  

This IA was validated in user testing several times over the course of the design.  Interestingly, the IA did not change that much from the 2nd iteration onward.

Design: Early concepts

Early concepts were preliminary explorations designed to get feedback on the structure and workflows / task flows.

In the interaction design phase we came up with some innovative features, such as the highly regarded Personal Profiles feature that simplifies setup and onboarding, as well as improves users’ understanding of how the pump is calculating insulin delivery.

Since there was no existing UI framework or API available to us, such as a mobile device platform like iOS or Android, I had to start thinking about this project as building an operating system from scratch and had to come up with a complete set of design patterns.

I designed navigation patterns, list views, grid lists, alert dialogs, info dialogs, animations and other UI patterns that were scaleable and essential to delivering a cohesive experience. 

I started building wireframes and screen layouts using the core design patterns and organized them into flows for each use case.

As each screen layout and flow became more mature I worked closely with visual design to polish the visuals with the appropriate UI patterns and mechanisms. 

While deciding on an interaction model, we wanted users to have immediate access to the bolus function.

User testing — formative

Throughout the design process, the team conducted in person interviews with a range of type 1 diabetes patients and nurse educators to evaluate the design.

I regularly worked with the researcher and helped conduct and moderate user testing. At times, I would tweak the prototype during the user testing session to pivot the task or get more info about a new idea.  

In formative user testing we found

  • There was too much information on the home screen and not enough focus.

  • The primary Orange color of the original interface was too bright and distracting

  • The Options screen was using grid lists rather than full list views, and some patients did not have the dexterity in their fingers to navigate a grid list as they did a regular list view.

  • User did not need the ability to change their basal insulin setting as often as initially thought.  As such, we decided to remove insulin delivery profiles from the home screen.

  • Extended bolus (extending an infusion over the coarse of time) concept was not understood using the toggle switch showing each toggle state. We later changed the switch to ‘On/Off’ and labeled it ‘Extended’ and somehow this lockup tested very well.

  • The ability to set a profile with a group of settings appropriate for a specific activity by allowing them to be customized and saved varations of their main profile was the favorite feature in testing.

So, we went back and updated the design

  • Cleaned up the home screen and removed unnecessary information

  • Added persistent insulin on board information on the home screen

  • Introduced full list views to replace the grid lists for the Options menu

  • Expanded personal profiles which streamline setup and adjusting basal insulin while doing something active

  • Reworked the extended bolus flow

  • Reworked the Stop/Resume Insulin feature

  • Introduced a screen lock mechanism

  • Enhanced the visual design with a dark color scheme using bright, high contrast colors

Design: Updated concepts

  • Cleaned up the home screen and removed unnecessary information

  • Intentionally focused the colors of the screen based on various factors such as color blindness, sunlight, display brightness, battery consumption, etc.

  • Added persistent insulin on board information on the home screen

  • Introduced full list views to replace the grid lists for the Options menu

  • Expanded personal profiles which streamline setup and adjusting basal insulin while doing something active

  • Reworked the extended bolus flow

  • Reworked the Stop/Resume Insulin feature

  • Introduced a screen lock mechanism

  • Enhanced the visual design with a dark color scheme using bright, high contrast colors

Prototypes

  • After each design iteration we would build a prototype of the feature to evaluate in user testing. Both paper prototypes and interactive click through prototypes were used to test the screens and flow.

    HTML, Flash and real working pump prototypes were built for evaluation.

Through iteration and user feedback, we solved 100% of the critical errors identified in previous iterations.

User testing — summative

In summative user testing we found

  • In version 4.1 we had over 100+ critical errors, with 9 unique errors

  • In version 4.2 we had 70 critical errors, with 5 unique errors

  • In version 4.3 there were 0 critical errors and 0 unique errors.

As shown in the slides below, we measured: Average errors per task, average task times, average ease of use.

FDA regulatory approval

As a regulated industry and product, I worked closely with the FDA to ensure our design documentation was thorough, complete, and clear in defining the product we were submitting for regulatory approval.