Reclaiming doctors’ time with AI-assisted rota management

Reclaiming doctors’ time with AI-assisted rota management

Reclaiming doctors’ time with AI-assisted rota management

​​RotaWiz is a proof-of-concept native app that unifies fragmented hospital rotas, automates safe shift swaps with AI, and exports schedules to personal calendars — helping doctors complete key tasks up to 96% faster and reduce cognitive load.
​​RotaWiz is a proof-of-concept native app that unifies fragmented hospital rotas, automates safe shift swaps with AI, and exports schedules to personal calendars — helping doctors complete key tasks up to 96% faster and reduce cognitive load.

​​RotaWiz is a proof-of-concept native app that unifies fragmented hospital rotas, automates safe shift swaps with AI, and exports schedules to personal calendars — helping doctors complete key tasks up to 96% faster and reduce cognitive load.

Role

Role

Sole end-to-end UX/UI designer

Sole end-to-end UX/UI designer

Industry

Industry

Healthcare SaaS

Healthcare SaaS

Time Frame

Time Frame

6 weeks

6 weeks

Background

Doctors perform a life-saving role, with life-draining admin

Doctors perform a life-saving role, with life-draining admin

Doctors perform a life-saving role, with life-draining admin

In the UK, hospital rotas are often scattered across spreadsheets, intranets, rota apps, paper printouts, and even WhatsApp threads. Versions multiply, updates go unseen, and doctors are left second-guessing where they’re meant to be and when. Swapping a shift adds another layer of friction — a slow, manual chain of messages, checks, and approvals that drain energy and invite error.

Challenge
Challenge
Challenge

How might we design a rota app MVP that integrates with hospital systems, ensures safety compliance, and gives clinicians one reliable, offline-accessible place to manage and swap shifts, helping them to reduce unpaid admin time and cognitive load?

Solution
Solution
Solution

RotaWiz — a native app MVP that:

  • Pulls a clinician’s 9–5 and on-call shifts into one unified, offline-capable view

  • Uses AI to suggest safe, policy-compliant shift swaps and handle requests

  • Exports confirmed rotas to personal calendars for easy sharing

Competitor analysis

How do doctors feel about existing rota software?

How do doctors feel about existing rota software?

How do doctors feel about existing rota software?

Doctors are among the most time-pressured professionals, yet much of their daily workload is weighed down by fragmented IT systems and unnecessary admin, from chasing the latest rota to securing time off. Dedicated medical rota software exists, but adoption and satisfaction remain low. I wanted to understand why.

Before speaking to users, I conducted an in-depth analysis of five popular medical rota software solutions.

Patchwork Health

Loop by RLDatix

HealthRota

MediRota

RotaPal

On paper, these all looked like comprehensive, modern tools. In reality, app store reviews told stories of clunky processes, missing features and frustrating interactions like frozen screens or duplicate content. It became clear that the issue wasn’t a lack of solutions, rather a lack of learnability, relevance and usability.

Before speaking to users, I conducted an in-depth analysis of five popular medical rota software solutions.

On paper, these all looked like comprehensive, modern tools. In reality, app store reviews –and later, user interviews– told stories of clunky processes, missing features and frustrting interactions. It became clear that the issue wasn’t a lack of solutions, rather a lack of learnability, relevance and usability.

User research

To understand how clinicians manage their schedules and where frustration peaks, I interviewed six senior UK doctors working in general medical specialties.

To understand how clinicians manage their schedules and where frustration peaks, I interviewed six senior UK doctors working in general medical specialties.

To understand how clinicians manage their schedules and where frustration peaks, I interviewed six senior UK doctors working in general medical specialties.

My goal was to find out:

My goal was to find out:

How are rotas currently distributed and accessed?

How are rotas currently distributed and accessed?

How are shift swaps and leave requests generally handled?

How are shift swaps and leave requests generally handled?

What personal workarounds are doctors implementing?

What personal workarounds are doctors implementing?

What is the impact of these fragmented systems on clinical operations?

What is the impact of these fragmented systems on clinical operations?

What is actually required to make a more successful system for all?

What is actually required to make a more successful system for all?

Key insight 1

Rotas are spread across multiple different formats, leading to errors, frustration and uncertainty

Rotas are spread across multiple different formats, leading to errors, frustration and uncertainty

Rotas are spread across multiple different formats, leading to errors, frustration and uncertainty

Every doctor described the same setup: separate 9–5 and on-call rotas — each created by different people, in different formats, stored in different locations, and updated unpredictably. The on-call rota often existed in multiple conflicting versions.

An example of a typical on-call (a) and 9-5 (b) hospital rota; one in spreadsheet form, the other on paper. Identifiable information has been redacted for confidentiality.

While some NHS trusts use dedicated rota software, many rely on a patchwork of complex spreadsheets and paper printouts pinned to the walls of common rooms. The result is a system that doctors consistently described as fragmented, inconsistent, and difficult to trust.

To cope, doctors develop their own workarounds — triple-checking details, screenshotting rotas for offline access, and manually adding shifts to personal calendars. But these patches come at the expense of time and mental energy.

Key insight 2

Swapping a shift is slow, admin-heavy, and prone to operational hazards

Swapping a shift is slow, admin-heavy, and prone to operational hazards

Swapping a shift is slow, admin-heavy, and prone to operational hazards

Swapping a shift to accommodate personal plans emerged as one of the biggest pain points; a process that took significant time and thought, often spread across multiple days. Users commonly described it as “time-consuming,” “awkward,” and “a nightmare”.

Because each swap request requires manual validation, the rota coordinators who create the schedules often fall behind on approvals and updates, and this risks cascading into clinical hazards. Even when swaps are approved, switchboards or contact lists for a given shift aren’t reliably updated, leading to understaffing, confusion or crisis mid-shift.

A typical text exchange between clinicians trying to swap a shift. Names have been redacted for confidentiality.

Research synthesis

Bringing focus to what matters most in everyday rota management

Bringing focus to what matters most in everyday rota management

Bringing focus to what matters most in everyday rota management

Interview findings were first grounded in a representative registrar persona to reflect common constraints and behaviours across participants. An empathy map was then used to translate those insights into daily pain points, decision pressures, and unmet needs, helping surface the top tasks that would later shape the MVP.

Ideation & prioritisation

I saw a clear opportunity to unify information and use AI to automate safe decision-making, helping doctors manage their shifts efficiently, and within policy rules.

I saw a clear opportunity to unify information and use AI to automate safe decision-making, helping doctors manage their shifts efficiently, and within policy rules.

I saw a clear opportunity to unify information and use AI to automate safe decision-making, helping doctors manage their shifts efficiently, and within policy rules.

For hospitals, fewer rota errors meant fewer coverage gaps, less admin, and reduced risk — a direct operational benefit.

In practice, new rota software often launches first on web and mobile platforms, with native apps introduced later once core systems are in place. For this project, I focused on the reverse: designing a proof-of-concept native app; a blueprint for how the experience could work once integrated with existing hospital systems.

How can AI increase efficiency – without stealing anyone's job?

How can AI increase efficiency – without stealing anyone's job?

How can AI increase efficiency – without stealing anyone's job?

I set realistic boundaries for scope. If the app was to be launched, rota coordinators would continue to create and manage rotas, but the app would automatically sync and notify users of changes, removing the need for manual updates, emails, and approvals.

From a business standpoint, the aim was to cut uneccessary admin time, build adoption, and reduce reliance on temporary locum staff caused by rota gaps — saving NHS trusts time and money, and improving morale across multidiscipliniary teams.

Time constraints meant it wasn't feasible to tackle every problem at once

Time constraints meant it wasn't feasible to tackle every problem at once

Time constraints meant it wasn't feasible to tackle every problem at once

User research had revealed four major pain points:

User research had revealed four major pain points:

User research had revealed four major pain points:

Viewing shift details via a unified rota

Swapping shifts through AI assisted matching

Requesting annual or study leave

Exporting the rota to a personal calendar

To stay focused, I brainstormed potential solutions and used the MoSCoW matrix to separate what the MVP must deliver from what could follow later. This helped ensure the app addressed the highest-impact problems first.

While requesting annual leave was a significant source of frustration for many, it was also the most technically complex to tackle, as it hinged on setting up a complex series of rules that wouldn't be feasible to design for in the time available.

Since accessing the rota and swapping shifts are more frequent tasks than requesting leave, I hypothesised that streamlining these processes first would already go a long way to reducing unpaid time and cognitive load.

So for the MVP, I focused in on three key flows:

So for the MVP, I focused in on three key flows:

So for the MVP, I focused in on three key flows:

Viewing shift details via a unified rota

Swapping shifts through AI assisted matching

Exporting the rota to a personal calendar

Must-have features focused on introducing AI, integrating with existing apps and progressively disclosing information

Must-have features focused on introducing AI, integrating with existing apps and progressively disclosing information

Must-have features focused on introducing AI, integrating with existing apps and progressively disclosing information

Unified rota

Combines daytime and on-call shifts into a single calendar, reducing the need to manually cross-reference two or more spreadsheets / formats.

Daily, weekly & monthly calendar views

Monthly view to scan shifts at a glance, weekly view for expanded detail, daily view for full detail + further actions.

Colour-coded shift types

e.g. night, late, twilight, annual leave, to allow for rapid scanning of the calendar.

Shift location and role tagging

Each rota entry includes site, ward/clinic/theatre and role (e.g. ward cover, acute take).

Colleague visibility

Shows colleagues on the same shift, including their role and grade.

Personal calendar integration

Export shifts to iCal, Google Calendar and Microsoft Outlook.

AI-assisted shift swapping

Suggests valid swap options based on rota compliance rules (e.g. rest periods, max nights in a row), ranks best options, and notifies potential swap partners.

Designing in the absence of a clear benchmark

Designing in the absence of a clear benchmark

Because existing rota apps are locked to hospital networks, I couldn’t study their designs or interactions directly. With few comparable products to benchmark, I explored adjacent patterns instead, analysing how calendar and shift management apps reveal event details, introduce AI, and handle calendar exports.

These insights informed early wireframes that felt both familiar and contextually appropriate for the healthcare environment.

Mid fidelity usability testing

A/B testing layout variations to ensure dense rota data was simple and intuitive

A/B testing layout variations to ensure dense rota data was simple and intuitive

A/B testing layout variations to ensure dense rota data was simple and intuitive

For the unified rota, I needed to find a way to clearly show different shift types without resorting to the visual overwhelm inherent in spreadsheets. I explored widely on paper before settling on two rota layouts to A/B test with users.

The majority of users preferred the scrollable carousel calendar and the spacious list view, so I took these forward into the mid-fidelity usability test.

Testing usability, relevance and cognitive ease in mid-fidelity

Testing usability, relevance and cognitive ease in mid-fidelity

Testing usability, relevance and cognitive ease in mid-fidelity

Using the mid-fidelity prototype, I ran remote moderated usability sessions with 5 senior UK doctors.

Concepts I tested

Concepts I tested

Viewing the rota

Viewing the rota

Could users easily locate, filter, and interpret their shifts?

Could users easily locate, filter, and interpret their shifts?

Requesting a shift swap

Requesting a shift swap

Did the AI-assisted swap flow feel realistic, efficient, and trustworthy? 

Did the AI-assisted swap flow feel realistic, efficient, and trustworthy? 

Exporting the rota to a personal calendar

Exporting the rota to a personal calendar

Could users confidently find and trust this task?

Could users confidently find and trust this task?

The solution was validated, with 100% task success rate across flows…

The solution was validated, with 100% task success rate across flows…

The solution was validated, with 100% task success rate across flows…

5/5 users felt the app reduces mental burden compared to existing rota systems

Users described the app as "clear", "streamlined" and "well thought-through"

The AI swap assistant was universally seen as time-saving, intuitive and a major improvement

Iterations round 1

But minor tweaks focused on realism and discoverability were required

But minor tweaks focused on realism and discoverability were required

But minor tweaks focused on realism and discoverability were required

Although no issues with the actual flows came up, users helped me identify areas where the content could be made more representative and useful for the contexts they would experience in real life.

Key changes from mid-fi to hi-fi

Key changes from mid-fi to hi-fi

Key changes from mid-fi to hi-fi

Some users wanted to be able to see their colleague's rota. When clicking on a colleague on the same shift, users can now click between tabs showing their colleague's contact information and their rota.

I replaced early shift types with twilight shifts across the board, as many noted that early shifts were rare, with some never even having worked one.

A couple of users found the export icon ambiguous, so I replaced it with a clearly labelled floating action bar above the rota.

Most felt that the window for being able to accept a swap request was too long, so I shortened it to 48h with an automatic system nudge thereafter.

Hi-fidelity usability testing

RotaWiz cut average task time by more than 85%

RotaWiz cut average task time by more than 85%

RotaWiz cut average task time by more than 85%

*Average based on self reported time taken to manually complete tasks.

Users described the concept as intuitive, streamlined and time-saving

Users described the concept as intuitive, streamlined and time-saving

Users described the concept as intuitive, streamlined and time-saving

Creating a design system

Auditing UI to go from styles —> scalable design system

Auditing UI to go from styles —> scalable design system

Auditing UI to go from styles —> scalable design system

When first establishing the visual direction, I'd set up some basic text, elevation and colour styles.

When first establishing the visual direction, I'd set up some basic text, elevation and colour styles.

When first establishing the visual direction, I'd set up some basic text, elevation and colour styles.

I used navy as the primary for its healthcare connotations of safety, with vibrant teal for actions. The data palette was trickier — I needed to distinguish multiple shift types without creating visual overwhelm. I landed on pastels in adjacent hues as a solution, as is seen in many calendar apps.

For elevation, I set S, M and L styles, shifting naturally from close and harsh to soft and spread. 

For typography, I wanted clean and professional with a hint of personality — a geometric sans serif felt right, and after testing several I landed on Helvetica Neue, using weight and size alone to create hierarchy and balance the busy colour palette. 

I used navy as the primary for its healthcare connotations of safety, with vibrant teal for actions. The data palette was trickier — I needed to distinguish multiple shift types without creating visual overwhelm. I landed on pastels in adjacent hues as a solution, as is seen in many calendar apps.

For elevation, I set S, M and L styles, shifting naturally from close and harsh to soft and spread. 

For typography, I wanted clean and professional with a hint of personality — a geometric sans serif felt right, and after testing several I landed on Helvetica Neue, using weight and size alone to create hierarchy and balance the busy colour palette. 

But following user testing, I noticed that several instances of inconsistency had snuck in – and the limited styles I had weren't well set up to scale.

But following user testing, I noticed that several instances of inconsistency had snuck in – and the limited styles I had weren't well set up to scale.

But following user testing, I noticed that several instances of inconsistency had snuck in – and the limited styles I had weren't well set up to scale.

Starting with an audit of every colour, font size, weight and spacing value in use, I rebuilt the system from scratch.

I first built out colour scales with tints and shades across the brand, data, semantic and neutral colour categories, giving flexibility across states and modes. Each scale was tested for accessibility; minimum AA was the baseline, but around 90% of colours meet AAA.

Primitive tokens were then mapped to semantic tokens for surfaces, backgrounds, borders, text and so on. I repeated this process for typography, radius and spacing, setting up the respective primitive and semantic tokens. 

The "sun" and "purple" data colours, used to illustrate anything in the app related to a long or twilight shift respectively, illustrates this approach below.

Starting with an audit of every colour, font size, weight and spacing value in use, I rebuilt the system from scratch.

I first built out colour scales with tints and shades across the primary, secondary, data, neutral and semantic colour categories, giving flexibility across states and modes. Each scale was tested for accessibility; minimum AA was the baseline, but around 90% of colours meet AAA.

Primitive tokens were then mapped to semantic tokens for surfaces, backgrounds, borders, text and so on. I repeated this process for typography, radius and spacing.

The "sun" and "purple" data colours, used to illustrate anything in the app related to a long or twilight shift respectively, illustrates this approach below.

Grounding the design system in atomic design principles

Grounding the design system in atomic design principles

Finally, I rebuilt the components, focusing on an atomic design system where my base components were atoms that could be used to build molecules and organisms – the larger and more complex components.

Molecules were designed with component properties to maximise flexibility, with elements like icons and badges swappable or hidden depending on context. By exposing these as nested instances within the organisms, I created a highly adaptable and customisable component set — extended further through variants.

Finally, I rebuilt the components, focusing on an atomic design system where my base components were atoms that could be used to build molecules and organisms – the larger and more complex components.

Molecules were designed with component properties to maximise flexibility, with elements like icons and badges swappable or hidden depending on context. By exposing these as nested instances within the organisms, I created a highly adaptable and customisable component set — extended further through variants.

  1. Applying tokens and variables to atoms…

Applying tokens and variables to atoms…

  1. Which combine to create molecules…

Which combine to create molecules…

  1. Exposing nested instances to create flexible organisms…

Exposing nested instances to create flexible organisms…

  1. …And templates.

…And templates.

Iterations round 2

Following hi-fi testing, there were just a few final tweaks to make

Following hi-fi testing, there were just a few final tweaks to make

Following hi-fi testing, there were just a few final tweaks to make

Before

Before

Before

Users felt discomfort at the prospect of contacting colleagues they didn’t know to swap a shift, and noted that there should usually be more than 3 people available to swap with

Users felt discomfort at the prospect of contacting colleagues they didn’t know to swap a shift, and noted that there should usually be more than 3 people available to swap with

After

After

After

Users can now select their "preferred colleagues" the first time they click to swap shift

Users can now select their "preferred colleagues" the first time they click to swap shift

Before

Before

Before

Colleague details didn't exactly contain the content or placement users would expect
Colleague details didn't exactly contain the content or placement users would expect

Colleague details didn't exactly contain the content or placement users would expect

After

After

After

Updated colleague details to better fit doctor's mental models and real-life use cases

Updated colleague details to better fit doctor's mental models and real-life use cases

Final solution

Final solution

Final solution

A unified, AI-assisted rota experience that brings scattered schedules into one reliable view, automates safe shift swaps, and cuts key admin tasks by up to 96%—reducing cognitive load and giving clinicians back time to focus on patient care and recovery at work, and personal recovery outside of it.

Next project

Next project

Next project

Re-designing symptom logging for casual and power users at Flo Health

Re-designing symptom logging for casual and power users at Flo Health

Re-designing symptom logging for casual and power users at Flo Health

Femtech

Product strategy

User re-engagement

Conversion optimisation