SafeRide
The SafeRide Project Case Study was developed through ELVTR’s AI for UX Designers program to expand expertise and deepen leadership in AI‑driven design.
The program brought together the full range of skills and tools explored throughout the course, demonstrating strategic AI integration at every stage of the UX design process.
SafeRide
PROMPTING FOR IDIATION
5/22
Below are the AI tools used to implement this project.
The SafeRide Project Case Study was developed through ELVTR’s AI for UX Designers program to expand expertise and deepen leadership in AI‑driven design.
The program brought together the full range of skills and tools explored throughout the course, demonstrating strategic AI integration at every stage of the UX design process.
TARGET AUDIENCE
2/15
-
Parents who want reassurance and visibility into their child’s commute
-
Students grades 5+ who need simple and reliable ways to get to school or extracurricular activities
-
Adult college students looking for safe, efficient, and affordable transportation options
Quantitative Goals
4/15
Potential Impacts
User Growth Retention
5,000
active users within the first 6 months of launch.
Operational Efficiency
15 - 20%
reduction in missed or late pickups.
Engagement
70%
of users interacting with SafeBot at least once per week.
Safety and Trust
100%
compliance with Data Protection standards and regulations.
ChatGPT Reasonings
Aligns with what schools or local communities can realistically achieve in a controlled rollout.
The targets were set to reflect realistic efficiency gains achievable through automation.
Metrics were chosen to reflect healthy engagement without app fatigue.
Align with industry norms and support a mission-driven focus on safety and transparency.
Overview
1/15
Duration:
August 18 - November 3 (12 weeks)
Role:
UX Design Lead
Responsibilities:
Led end-to-end UX design for a ride-share transportation platform, leveraging AI tools for research, interaction design, prototyping, usability testing, and AI-driven feature development.
Description:
SafeRide is a mobile and desktop application created to support the education and transportation sectors.
It provides users with safe and easy access to bus routes and stop information, while integrating with community rideshare groups.
Included as part of its concept is an AI component entitled SafeBot – an AI-powered virtual assistant designed to help users navigate transportation options in addition to simplifying and safeguarding daily transport operations.
GOALS
The ChatGPT platform was used to evaluate potential impacts based on industry standards. Starting with the problem statement and project description as the initial prompt, I gradually added additional requirements, building on each prompt to refine the outputs. This iterative process helped surface the quantitative and qualitative goals outlined in this section.
QUALATATIVE GOALS
5/15
Operational Efficiency
Enhance the daily transportation experience, Increase punctuality and overall efficiency for students and families
Safety & Trust
Reduce stress and safety concerns for users
Engagement
Foster safe, communicative relationships within communities through the platform
User Growth Retention
Enable customizations real-time safety alerts and updates
PROMPTING
During early concept development, I explored how AI-supported ideation could accelerate branding and product positioning. ChatGPT and IBM Consulting Advantage tools were used to support early concept development.
PROMPTING EXAMPLES FOR DESIGN
5/15
Microsoft Copilot was used to generate initial logo concepts. Leveraging its variations, I then refined and developed the direction into a cohesive, visually compelling design aligned with the final color theme.
AI Generated Logo Variations
Final Design
PROMPTING FOR IDIATION
5/15
Guided by the project’s focus on safe and efficient transportation for students, ChatGPT and IBM Consulting Advantage, tools were used to support early concept development.
Displayed is the initial prompt that served as the foundation for the entire project. Using a consistent prompt enabled comparative evaluation of outputs and workflows across multiple AI platforms.
Throughout the design process, I continuously built upon this prompt, expanding it with additional questions to guide design decisions.
Case Study Prompt input
Prompt
You are a UX Designer. Create a mobile and responsive web wireframe for an app called SafeRide. The app helps parents, K-12 students, and adult college students find bus routes and stop information. Include features for searching and creating carpooling groups. Highlight safety with real-time student/child journey tracking. Integrate an Al chatbot assistant for route info, carpool organization, safety updates, and onboarding support. The design should be simple, intuitive, and community-focused.
Discovery and Research strategy
5/15
Core qualitative survey questions were developed and AI-assisted support was used to expand the survey with targeted multiple-choice and open-ended questions.
Key Insights
71.4%
Drive themselves or their children to extracurricular activities.
42.9%
Would be interested in using a chatbot to help find or organize a carpool group for themselves, a student, or child?
71.4%
Find it important to receive real-time updates (via chatbot) about a student's bus or carpool location?
6 out of 7
Find it helpful to have a chatbot that could answer questions about bus schedules and stop locations.
Discovery and Research strategy
5/15
The goal was to evaluate the overall product concept and the SafeBot feature, validate key assumptions, and surface user priorities and unmet needs. The survey was distributed and analyzed using Google Forms, with seven participants contributing responses.
Sentiment Analysis
Likes
Most participants are parents who seem to prefer “driving themselves” to their activities
It’s more “convenient” and feel they have “more control”
Dislikes
Some participants “dislike traffic”
Participants dislike “using their own gas”
It’s more “time consuming”
AI Solution
Participants found that an AI chatbot for the ride share app would “make life easier”
It would be “helpful for students because they won’t be late” and they can “save money on cabs”
Suggestions
Participants feel that the app would help them “feel that their child is safer”
Important features are to have “background checks of drivers”
Parents setting up their own rideshare group and “access to their child’s locations”
Design
The SafeRide design approach integrates AI as a strategic partner throughout the design development process. AI tools such as ChatGPT, IBM ICA, FontJoy and Coolors, were used to generate ideas for conversational flows, understand natural language patterns, and identify potential use cases for the SafeBot feature.
Competitive Research
5/15
Competitive research included a review of industry articles and company resources, including Medium and Uber. AI-assisted analysis was used to help identify best practices and actionable insights which informed feature and design recommendations for the SafeRide app.
Prompt Input
Prompt
Based on the article links provided below, what three recommendations and insights should I consider for enhancing the SafeRide app. Provide three recommendations for each.
AI Output Recommendations and Practices
-
In SafeRide, use data to suggest optimal bus routes or carpool matches automatically.
-
Display real-time tracking of buses or carpool vehicles, update estimated arrival times and trips. Notify users of deviations or delays.
-
Simplify and rationalize reporting options by use of color and visual hierarchy.
-
For safety-critical flows such as tracking route, confirming child arrival, raising alerts, streamline the process to 1–2 taps so users don’t get lost in menus.
Chatbot Design Narrative
5/15
User Flow Diagram
Pictured is the SafeRide user flow diagram for the chatbot feature outlining the step‑by‑step path a user would take when interacting with the SafeBot (Chatbot) feature. It begins on the home screen, where the user selects the SafeBot option, then moves into a guided conversational interface.
Persona Defined
5/15
Meet Ava:
Ava is a 23-year-old graduate student who works as a part-time adjunct instructor at the university she attends.
User Scenario:
Ava recently moved to an apartment 2.5 miles off campus. While she still enjoys daytime walks on campus, her new location and busy schedule make it harder to do so regularly. In the evenings, Ava prefers using rideshare services and feels safer riding with fellow students.
Expectations:
-
Receive updates on available drivers quickly and easily
-
Feel confident knowing safety standards are always met.
-
Count on timely pick-ups and arrivals — because being reliable is part of being socially responsible.
User Journey
5/15
The SafeRide user journey map outlines the end‑to‑end experience of Ava as she interacts with the app as a first time user. It begins with logging in, exploring key features to search for rides, accessing the SafeBot feature and finally accepting a ride.
AI Tone and Language Recommendations
5/15
I have determined that setting a friendly, accessible, and calm tone will make SafeRide feel trustworthy for both students and parents, especially in safety or emergency contexts.
Prompt Input
Prompt
I'm designing a SafeRide app for students, adult college student and parents and want the tone of the app to be friendly, accessible, and appropriate for emergencies. It should communicate clearly at a fifth-grade reading level. Can you generate sample chatbot responses or Ul text that reflect this tone? Provide other suggestions as well.
AI Output
Tone and Language Examples
5/15
The tone and language is intentionally simple and friendly. Messages are designed to be concise yet warm, helping users feel confident and supported throughout their experience.
Everyday Use
Find me a route for school.
Sure thing! The fastest route to Maple Middle School leaves at 7:15 a.m. from Pine Street. Want me to send you an alert 10 minutes before pickup?
Where is my ride now?
Your bus is on the way! It’s two stops away and should arrive in about 6 minutes.
I need to join a carpool.
Got it! I found two nearby carpool groups heading to your campus. Would you like me to show them?
Safety and Emergency Use
I feel unsafe walking home.
I’m here for you. I can share your location with your emergency contact or suggest a safer route. Which would you like me to do first?
Where is my ride now?
No worries—it happens! I can help you find another route or message your parent or driver. What would you like to do?
Help, I need emergency assistance!
I’m sending an alert to your emergency contact now. Stay where you are if it’s safe. I’ve also marked your location for responders
PROMPTING EXAMPLES for Ideation
5/15
Prompt input
Prompt
Referencing the description, generate practical names for the ride share app.
The app is for students and parents of students.
ChatGPT Output
I explored ChatGPT to generate app names and taglines. landed on SafeRide which were a combination of 2 names listed.
The tagline it “Smarter Routes, Safer Journeys”
PROMPTING EXAMPLES FOR DESIGN
5/15
To establish a cohesive visual identity for the SafeRide app Fontjoy was used to quickly explore typography combinations without manually pairing dozens of typefaces.
The Coolors platform was used to develop a palette that communicated safety, clarity, and community. This process ensured the final color system met WCAG contrast standards.
FontJoy: Montserrat font
Coolors: Curated Color Palette
Figma Make
5/15
The Problem
5/15
Key Insights
Many families experience daily uncertainty and frustration due to unreliable school transportation systems. Inconsistent bus arrivals, lack of real-time communication, and limited accountability create anxiety for parents and disrupt students’ education.
Problem statement
Parents and students need a reliable, transparent way to track transportation and coordinate safe alternatives because current systems lack visibility, communication, and trust.
Prototype
The ChatGPT platform was used to evaluate potential impacts based on industry standards. Starting with the problem statement and project description as the initial prompt, I gradually added additional requirements, building on each prompt to refine the outputs. This iterative process helped surface the quantitative and qualitative goals outlined in this section.
Overview
Duration:
August 18 - November 3 (12 weeks)
Role:
UX Design Lead
Responsibilities:
Led end-to-end UX design for a ride-share transportation platform, leveraging AI tools for research, interaction design, prototyping, usability testing, and AI-driven feature development.
Description:
SafeRide is a mobile and desktop application created to support the education and transportation sectors.
It provides users with safe and easy access to bus routes and stop information, while integrating with community rideshare groups.
Included as part of its concept is an AI component entitled SafeBot – an AI-powered virtual assistant designed to help users navigate transportation options in addition to simplifying and safeguarding daily transport operations.
Context
Prompt Input and AI Tools
Displayed is the initial prompt that served as the foundation for the entire project.
Throughout the design process, I continuously built upon this prompt, expanding it with additional questions to guide design decisions.
Case Study Prompt input
Prompt
You are a UX Designer. Create a mobile and responsive web wireframe for an app called SafeRide. The app helps parents, K-12 students, and adult college students find bus routes and stop information. Include features for searching and creating carpooling groups. Highlight safety with real-time student/child journey tracking. Integrate an Al chatbot assistant for route info, carpool organization, safety updates, and onboarding support. The design should be simple, intuitive, and community-focused.
Using a consistent prompt enabled comparative evaluation of outputs across multiple AI platforms.
Empathize
1
Project Challenge
Problem Statement
Parents and students need a reliable, transparent way to track transportation and coordinate safe alternatives.
Key Painpoints
Unreliable Transportation
Families experience stress due to unreliable school transportation systems and unpredictable bus schedules.
Lack of Trust
There is low trust and accountability and limited coordination between schools, parents, and transportation providers.
Lack of Communication
There is a lack of real-time communication and scheduling updates.
Target Audience
-
Parents who want reassurance and visibility into their child’s commute
-
Students grades 5+ who need simple and reliable ways to get to school or extracurricular activities
-
Adult college students looking for safe, efficient, and affordable transportation options
2
Discovery and Research
User Research
Initial survey questions were developed and AI-assisted support was used to evaluate the overall product concept and the SafeBot feature, validate key assumptions, and surface user priorities and unmet needs.
The survey was distributed and analyzed using Google Forms, with seven participants contributing responses.
Key Insights
71.4%
Drive themselves or their children to extracurricular activities.
42.9%
Would be interested in using a chatbot to help find or organize a carpool group for themselves, a student, or child?
6 out of 7
Find it helpful to have a chatbot that could answer questions about bus schedules and stop locations.
71.4%
Find it important to receive real-time updates (via chatbot) about a student's bus or carpool location?
The goal was to evaluate the overall product concept and the SafeBot feature, validate key assumptions, and surface user priorities and unmet needs. The survey was distributed and analyzed using Google Forms, with seven participants contributing responses.
User Feedback
Likes
Most participants are parents who seem to prefer “driving themselves” to their activities
Participants feel the app would be more “convenient”, they'll have “more control” and would “feel that their child is safer”
Dislikes
Some participants “dislike traffic”, It’s more “time consuming”
Participants dislike “using their own gas”
AI Solution
Participants found that an AI chatbot for the ride share app would “make life easier”
It would be “helpful for students because they won’t be late” and they can “save money on cabs”
Suggestions
Important features are to have “background checks of drivers”
Parents setting up their own rideshare group and “access to their child’s locations”
ChatGPT Reasonings
Define
3
Project Goals
Quantitive Goals with ChatGPT Reasonings
User Growth Retention
5,000
active users within the first 6 months of launch.
Aligns with what schools or local communities can realistically achieve in a controlled rollout.
Operational Efficiency
15 - 20%
reduction in missed or late pickups.
The targets were set to reflect realistic efficiency gains achievable through automation.
Engagement
70%
of users interacting with SafeBot at least once per week.
Metrics were chosen to reflect healthy engagement without app fatigue.
Safety and Trust
100%
compliance with Data Protection standards and regulations.
Align with industry norms and support a mission-driven focus on safety and transparency.
Qualitative Impacts with ChatGPT Reasonings
User Growth Retention
Enable customizations real-time safety alerts and updates
Operational Efficiency
Enhance the daily transportation experience, Increase punctuality and overall efficiency for students and families
Engagement
Foster safe, communicative relationships within communities through the platform
Safety & Trust
Reduce stress and safety concerns for users
The goal was to evaluate the overall product concept and the SafeBot feature, validate key assumptions, and surface user priorities and unmet needs. The survey was distributed and analyzed using Google Forms, with seven participants contributing responses.
4
User Definition
Persona
Meet Ava
Ava is a 23-year-old graduate student who works as a part-time adjunct instructor at the university she attends.
Expectations
-
Receive updates on available drivers quickly and easily
-
Feel confident knowing safety standards are always met
-
Count on timely pick-ups and arrivals
Scenario
Ava recently moved to an apartment 2.5 miles off campus. While she still enjoys daytime walks on campus, her new location and busy schedule make it harder to do so regularly. In the evenings, Ava prefers using rideshare services and feels safer riding with fellow students.
User Journey
Chatbot Design Narrative
Pictured is the SafeRide user flow diagram for the chatbot feature outlining the step‑by‑step path a user would take when interacting with the SafeBot (Chatbot) feature. It begins on the home screen, where the user selects the SafeBot option, then moves into a guided conversational interface.
SafeBot AI Chatbot User Flow Diagram
The user flow diagram maps how users interact with SafeBot to quickly access routes, coordinate rides, and receive real-time safety support.
AI tools FOR DESIGN
To establish a cohesive visual identity for the SafeRide app Fontjoy was used to quickly explore typography combinations without manually pairing dozens of typefaces.
The Coolors platform was used to develop a palette that communicated safety, clarity, and community. This process ensured the final color system met WCAG contrast standards.
FontJoy: Montserrat font
Coolors: Curated Color Palette
After evaluating each prototype, I identified key strengths across all tools and integrated them into a unified design.
What I Kept:
-
Simplicity and chatbot focus from Claude
-
Structured navigation from Figma Make
-
Visual hierarchy from UX Pilot
After evaluating each prototype, I identified key strengths across all tools and integrated them into a unified design.
What I Kept:
-
Simplicity and chatbot focus from Claude
-
Structured navigation from Figma Make
-
Visual hierarchy from UX Pilot
Ideate
3
AI Iterations
Rapid Prototyping Tools
Figma Make
Structured and Feature-Rich
Insights:
-
Clear navigation and multi-page structure
-
Strong feature visibility (Routes, Carpool, Tracking)
-
Felt more system-driven than user-friendly
UX Pilot
Simple & User-Friendly
Insights:
-
Good visual hierarchy
-
Balanced layout between features and usability
-
Strong foundation for scalable design system
Claude
Simple & User-Friendly
Insights:
-
Clean, minimal interface
-
Strong emphasis on chatbot interaction
-
More intuitive and approachable for younger users
Prototype
4
Design Direction
After evaluating each iteration, I identified key strengths across all tools and integrated them into a unified design. Pctured is the login page (1), home screen (2), and SafeBot assistant (3).
What I Kept:
-
Simplicity and chatbot focus from Claude
-
Structured navigation from Figma Make
-
Visual hierarchy from UX Pilot
Where is my ride now?
Your bus is on the way! It’s two stops away and should arrive in about 6 minutes.
I need to join a carpool.
Got it! I found two nearby carpool groups heading to your campus. Would you like me to show them?
Tone and Language
Prompt input
Prompt
I'm designing a SafeRide app for students, adult college student and parents and want the tone of the app to be friendly, accessible, and appropriate for emergencies. It should communicate clearly at a fifth-grade reading level. Can you generate sample chatbot responses or Ul text that reflect this tone? Provide other suggestions as well.
Everyday Use
Find me a route for school.
Sure thing! The fastest route to Maple Middle School leaves at 7:15 a.m. from Pine Street. Want me to send you an alert 10 minutes before pickup?
Safety and Emergency Use
I feel unsafe walking home.
I’m here for you. I can share your location with your emergency contact or suggest a safer route. Which would you like me to do first?
Help, I need emergency assistance!
No worries—it happens! I can help you find another route or message your parent or driver. What would you like to do?
I missed my bus.
I’m sending an alert to your emergency contact now. Stay where you are if it’s safe. I’ve also marked your location for responders
The tone and language is intentionally simple and friendly. Messages are designed to be concise yet warm, helping users feel confident and supported throughout their experience.
Branding and Color Palette
AI Generated Logo variations
Final Logo Design
FontJoy Font Pairings: Montserrat
Coolors Color Palette
#4F35EA
#DEE4FF
#FF6161
#EDEEF5
-
Using Microsoft Copilot, to generate initial logo concepts, I refined and developed the direction into a, visually compelling logo design using the color theme.
-
AI tools including FontJoy AI Font Generator and Coolors Color Palette Generator were used to explore font pairings and adjust color palette variations, with accessibility standards guiding final selections.
Test
5
User Testing
The success criteria for the user test require that users are able to navigate to the SafeBot assistant with clarity and ease.
Insights were gathered from a total of 8 participants.
What Worked Well
87.5% High Task Success Rate
A strong majority of participants were able to complete the primary task without assistance, indicating that the core user flow, navigation structure, and action hierarchy effectively support goal completion.
This suggests the interaction model is intuitive and aligned with user expectations.
Click below to explore the Maze SafeRide test link for this project.
What Didn't Work Well
71.0% Misclick Rate on the Home Screen
The image indicates that a significant number of participants selected incorrect elements when attempting to access the SafeBot feature.
This suggests that the feature’s placement, visual prominence, or labeling did not provide a strong enough visual cue, highlighting an opportunity to improve hierarchy and call-to-action clarity on the home screen.
Where Users Struggled
"If the SafeBot flow is the main flow then perhaps the button should be easier to spot or higher on the list."
Easy path to get to the SafeBot; however, when I got to the main menu, there was a lot of information for me to find SafeBot
"I was a little confused by the 4 main buttons on the home screen all being named SafeSomething so it took a second more to find the "Safebot tab"
Actionable Insights
User feedback reported difficulty distinguishing between features with similar naming conventions (e.g., SafeBot, SafeRide, and other “Safe–” labels) indicating that the naming system may create cognitive friction and reduce scannability.
Recommended Actions
-
Refine labeling for clearer differentiation and faster recognition.
-
Improve SafeBot visibility through visual hierarchy.
Synthetic Data Planning
A prompt was developed. Data was input into ChatGPT to generate the SafeRide synthetic data Excel document containing data sets simulating real-world user interactions.
Prompt Input
Prompt
Generate synthetic data for the following categories:
-
Date and Time
-
Scheduling
-
Availability
-
Geo Location
The "Scheduling" category includes the Class schedule, Activity schedule, Schedule of other household members and Bus schedule data.
The "Availability" category should include the availability of group members and the availability of group drivers.
The "Location" category should contain the home address and geo locations of Users.
Synthetic data was as developed to demonstrate how the app will manage data inputs and determine how AI will use the data to functionality support the app.
Mural Whiteboard
The use of Mural for synthetic data planning:
-
Organized and structured initial ideas
-
Identified datasets needed to support app functionality
-
Clarified how the data types influences component generation, design and behavior
Final Outcomes
6
AI Generated Possibilities
Transforming Synthetic Data Into UI Design
1
Using the prompt input below, the ChatGPT generated synthetic data Excel document was uploaded.
Prompt
Using the synthetic data, Create a series of components to integrate with my SafeRide app.
2
ChatGPT generated an initial prototype with React + Tailwind + Shadcn UI components integrated into a JavaScript file.
3
Components were refined in Claude uploading the JavaScript file.
Ethical and Data Privacy Solutions
An app that engages children and student data could create serious safety risks if not properly managed.
Therefore, ethical considerations were integrated throughout the design process, ensuring that SafeRide protects and respects sensitive assets.
1
Data Privacy & Security
-
User Consent
-
Data Encryption
-
Anonymized Testing Data
-
Data Minimization
2
Ethical AI Use
-
Transparency
-
Explainability
-
Fairness & Non-Discrimination
-
No Behavioral Manipulation
-
Continuous Monitoring
3
Accessibility & Inclusivity
-
Universal Design
-
Language Simplicity
-
Multi-interface Access
4
Data Governance
-
Parental Controls for minors
-
Data Retention Limits
-
Compliance
-
Third-Party Transparency
Home Screen Improvements
Based on usability findings, refined labeling was introduced to improve task clarity and reduce home screen misclicks.
Image 1: Initial Home Screen
Primary options were not visually differentiated, creating decision friction.
Image 2: Revised Home Screen
Clearer labeling and familiar iconography strengthen hierarchy, reduce cognitive load, and support faster user decision-making.
SafeRide App Design Improvements
1
Welcome: Introduces SafeRide with a clean, minimal interface that reduce friction for first-time users and reinforces trust and safety.
2
Login: Designed for clarity and ease of use, the login screen prioritizes accessibility with clear input fields, minimal distractions, and a straightforward path to account access.
3
Home: The home dashboard centralizes key features—rides, groups, journeys, and SafeBot—using clear visual hierarchy and iconography.
1. Welcome
2. Login
3. Home Page
4. SafeBot Default
5. AI Response
6. Map Integration
4
SafeBot Default State: The default chatbot state provides a welcoming entry point, guiding users to initiate interaction
5
AI Response: Shows a conversational interaction using natural language where users can request transportation assistance creating an intuitive user-friendly dialogue.
6
Map Integration: Introduces the SafeBot interface, combining conversational AI with location-based context.
Click below to view full Figma Prototype.
Please note that SafeRide is an ongoing project. The current case study highlights the core experience and initial UI direction, with additional features and components actively in development. I plan to continue refining and expanding the prototype through iterative testing and design improvements.
Lessons learned
AI is not a replacement.
it’s a collaborative partner that can accelerate the creative process, allowing more time for refinement and human-centered decision-making requiring governance.
AI tools can go beyond just answers to questions.
Instead of just providing information, AI tools demonstrates how designers can analyze where users might face friction and refine design concepts before testing begins.
There are benefits of synthetic data in design.
Synthetic data allowed realistic simulations of bus schedules, user locations, and group availability without compromising data privacy.
Accessibility is critical to inclusive design.
Accessibility considerations like contrast, readability, and inclusive language became central to making the chatbot usable for both students and parents.
The Mural whiteboard workspace was used to organize ideas and address ethical requirements and identify datasets that would address functional needs.
Additional components such as Class Schedule and Schedule of Household Members were required to implement the AI-driven features properly which will help to understand how the data directly influences component generation and design.
Continuing in ChatGPT, I entered a prompt which generated initial prototype components using JavaScript as the primary programming language (React + Tailwind + Shadcn/ui).
Pictured are screenshots of the components generated: (1) Bus Schedule, (2) Member availability, (3) Driver Availability, (4) Journey Events
Prompt
Using the synthetic data, create a series of components to integrate with my SafeRide App.
The prototype provided the option to upload the synthetic data where the SafeRide AI chatbot functionality (SafeBot) was able to be tested through inquiries.
In the chatbot demo, user inputs such as “Find me a bus route” triggered responses sourced directly from the available data.
Maze Analysis Prompt Input
The Maze report was imported into Copilot and analyzed using the following prompt:
Prompt
You are a Product Designer. Create a user testing report from the attached document. Please be brief.
Gather & Analyse Feedback:
-
Record what worked well and where users struggled.
-
Note any Al-generated insights (e.g., predicted attention hotspots, A/B test results).
Write a short reflection (5-7 sentences) on:
-
What you learned from user behavior.
-
How Al insights complemented or differed from human feedback.
-
What you would test next with more time.
Please include the correlating page numbers where you collected the information. Maze-Report_SafeRide Al Bot_9_29_2025.pdf
Usability Testing
The usability testing was conducted remotely using the Maze UX research and usability test platform. I imported the Figma prototype into the platform to gather insights to refine the app design, improve the overall experience, and ensure SafeRide is both user-friendly and trustworthy.
The success criteria for the user test require that users are able to navigate to the SafeBot assistant with clarity and ease.
Insights were gathered from a total of 8 participants.
Click below to explore the Maze SafeRide test link for this project.
Prototype development and Testing
First Iteration Test Prototype
Based on the comparative outputs, I've identified features that best met the project requirements to develop the first iteration of the test prototype. Guided by the user flow diagram, I created the prototype in Figma.
Users were asked to evaluate the usability of the Sign in and SafeBot feature in the SafeRide mobile app implementing the following tasks:
-
Sign in to the SafeRide application (1).
-
Navigate to the Safebot tab (2).
-
Type in a request asking the AI Bot to: “Help me find a but route to school” (3).
Claude Output
The Claude AI outputs shown below aligned most closely with the intended design direction. This variation presents a more approachable, youth-friendly experience, with particular strengths in the chatbot feature and the overall simplicity of the interface. These qualities informed key decisions carried forward into the initial prototype design.
Tone and Language Examples
Design Implementation
What Worked Well
87.5% High Task Success Rate
A strong majority of participants were able to complete the primary task without assistance, indicating that the core user flow, navigation structure, and action hierarchy effectively support goal completion.
This suggests the interaction model is intuitive and aligned with user expectations.
AI Bot Request Flow Felt Intuitive Once Discovered
After locating the SafeBot feature, users were able to move through the request process with minimal confusion. The images indicates no drop-offs and low misclicks on the AI response screens.
This indicates that the conversational structure, prompts, and step-by-step guidance successfully reduce cognitive load during task completion.
Positive Sentiment Toward Visual Design and Color Palette
Participants responded favorably to the overall interface aesthetic, particularly the blue color palette.
Users associated the colors with feelings of trust, calmness, and safety — emotional attributes that strongly support the app’s purpose in the context of student transportation.
This reinforces that the visual design is not only appealing, but also aligned with the product’s emotional and functional goals.
Where Users Struggled
High Misclick Rate on the Home Screen
The image indicates that a significant number of participants selected incorrect elements when attempting to access the SafeBot feature.
This suggests that the feature’s placement, visual prominence, or labeling did not provide a strong enough visual cue, highlighting an opportunity to improve hierarchy and call-to-action clarity on the home screen.
Confusion Between Similarly Named Features
User feedback reported difficulty distinguishing between features with similar naming conventions (e.g., SafeBot, SafeRide, and other “Safe–” labels).
This indicates that the naming system may create cognitive friction and reduce scannability, signaling a need to refine labeling for clearer differentiation and faster recognition.
Limited Visibility of the SafeBot Feature
Several participants also suggested making the SafeBot feature (1) more prominent on the home screen.
This feedback reinforces the need to elevate its visual priority through placement, contrast, iconography, or persistent access patterns.
AI Anaylsis of Insights
Conclusion of Analysis
In conclusion, the AI insights aligned with participant feedback, especially in identifying the home screen as a friction point. Among participants who responded, the sentiment scale average shows that 7 of the 8 participants rated the primary task easy to complete overall.
The high task success rate (100%) observed during testing suggests that the core user flow is intuitive once users are oriented within the interface.
While overall sentiment was positive, findings also highlighted opportunities to improve feature discoverability—particularly for the SafeBot entry point—which informed targeted design refinements for subsequent iterations.
Synthetic data was as developed to demonstrate how the app will manage data inputs and determine how AI will use the data to functionality support the app.
19
20
PROMPTING EXAMPLES FOR DESIGN
5/15
Final Iterations and Output
Home Screen Iteration
Based on usability findings, the first iteration focused on improving SafeBot visibility through hierarchy and refined labeling, improving task clarity to reduce home screen misclicks.
mage (1) shows the initial screen, where primary options were not visually differentiated, creating decision friction. In image (2), clearer labeling and familiar iconography were introduced to strengthen hierarchy, reduce cognitive load, and support faster user decision-making.
22
The Mural whiteboard workspace was used to organize ideas and address ethical requirements and identify datasets that would address functional needs.
Additional components such as Class Schedule and Schedule of Household Members were required to implement the AI-driven features properly which will help to understand how the data directly influences component generation and design.
21
A prompt was developed. The data was input into ChatGPT to generate the SafeRide synthetic data Excel document.
Prompt
Generate synthetic data for the following categories:
-
Date and Time
-
Scheduling
-
Availability
-
Geo Location
The "Scheduling" category includes the Class schedule, Activity schedule, Schedule of other household members and Bus schedule data.
The "Availability" category should include the availability of group members and the availability of group drivers.
The "Location" category should contain the home address and geo locations of Users.
20
Pictured is the synthetic data Excel document used to analyze and structure the SafeRide component feature development.
Functional UI Components
Continuing in ChatGPT, I entered a prompt which generated initial prototype components using JavaScript as the primary programming language (React + Tailwind + Shadcn/ui).
Pictured are screenshots of the components generated: (1) Bus Schedule, (2) Member availability, (3) Driver Availability, (4) Journey Events
Prompt
Using the synthetic data, create a series of components to integrate with my SafeRide App.
The prototype provided the option to upload the synthetic data where the SafeRide AI chatbot functionality (SafeBot) was able to be tested through inquiries.
In the chatbot demo, user inputs such as “Find me a bus route” triggered responses sourced directly from the available data.
The JavaScript file generated by ChatGPT was uploaded into Claude to refine the components The following prompt was entered.
Functional AI Chatbot
Refined Components in Claude
First Iteration Test Prototype
Tone and Language Examples
Key Insights
71.4%
Drive themselves or their children to extracurricular activities.
42.9%
Would be interested in using a chatbot to help find or organize a carpool group for themselves, a student, or child?
71.4%
Find it important to receive real-time updates (via chatbot) about a student's bus or carpool location?
6 out of 7
Find it helpful to have a chatbot that could answer questions about bus schedules and stop locations.
Likes
Most participants are parents who seem to prefer “driving themselves” to their activities
It’s more “convenient” and feel they have “more control”
Dislikes
Some participants “dislike traffic”
Participants dislike “using their own gas”
It’s more “time consuming”
AI Solution
Participants found that an AI chatbot for the ride share app would “make life easier”
It would be “helpful for students because they won’t be late” and they can “save money on cabs”
Suggestions
Participants feel that the app would help them “feel that their child is safer”
Important features are to have “background checks of drivers”
Parents setting up their own rideshare group and “access to their child’s locations”
First Iteration Test Prototype
Based on the comparative outputs, I've identified features that best met the project requirements to develop the first iteration of the test prototype. Guided by the user flow diagram, I created the prototype in Figma.
Users were asked to evaluate the usability of the Sign in and SafeBot feature in the SafeRide mobile app implementing the following tasks:
-
Sign in to the SafeRide application (1).
-
Navigate to the Safebot tab (2).
-
Type in a request asking the AI Bot to: “Help me find a but route to school” (3).
Maze Analysis Prompt Input
The Maze report was imported into Copilot and analyzed using the following prompt:
Prompt
You are a Product Designer. Create a user testing report from the attached document. Please be brief.
Gather & Analyse Feedback:
-
Record what worked well and where users struggled.
-
Note any Al-generated insights (e.g., predicted attention hotspots, A/B test results).
Write a short reflection (5-7 sentences) on:
-
What you learned from user behavior.
-
How Al insights complemented or differed from human feedback.
-
What you would test next with more time.
Please include the correlating page numbers where you collected the information. Maze-Report_SafeRide Al Bot_9_29_2025.pdf
Usability Testing
The usability testing was conducted remotely using the Maze UX research and usability test platform. I imported the Figma prototype into the platform to gather insights to refine the app design, improve the overall experience, and ensure SafeRide is both user-friendly and trustworthy.
The success criteria for the user test require that users are able to navigate to the SafeBot assistant with clarity and ease.
Insights were gathered from a total of 8 participants.
Click below to explore the Maze SafeRide test link for this project.
Prototype development and Testing
First Iteration Test Prototype
Based on the comparative outputs, I've identified features that best met the project requirements to develop the first iteration of the test prototype. Guided by the user flow diagram, I created the prototype in Figma.
Users were asked to evaluate the usability of the Sign in and SafeBot feature in the SafeRide mobile app implementing the following tasks:
-
Sign in to the SafeRide application (1).
-
Navigate to the Safebot tab (2).
-
Type in a request asking the AI Bot to: “Help me find a but route to school” (3).
Claude Output
The Claude AI outputs shown below aligned most closely with the intended design direction. This variation presents a more approachable, youth-friendly experience, with particular strengths in the chatbot feature and the overall simplicity of the interface. These qualities informed key decisions carried forward into the initial prototype design.
UX Pilot Output
The UX Pilot AI generated outputs shown below, rendered the prompt as a continuous single-page design, above the fold (1) and below the fold (2). While many of the buttons were nonfunctional placeholders, the UX Pilot output mirrored similar components generated by Figma Make such as Quick Access (A), Active Rides (B), Plan Your Ride (C), and Safety Features (D).
Figma Make Output
The Figma Make outputs shown below features the home page (1). Other AI generated pages are Routes (2), Carpool (3), Tracking (4), and AI Chat (5). Several Key features generated by Figma Make aligns well with the intended design direction for the SafeRide app which are quick actions, clearly visible active trip information, and the safety network among others.
Case Study Prompt Input
I expanded upon my prompt descriptions to guide early prototype generation, emphasizing key functional recommendations and app features such as journey tracking and real-time safety updates.
Case Study Prompt input
Prompt
You are a UX Designer. Create a mobile and responsive web wireframe for an app called SafeRide. The app helps parents, K-12 students, and adult college students find bus routes and stop information. Include features for searching and creating carpooling groups. Highlight safety with real-time student/child journey tracking. Integrate an Al chatbot assistant for route info, carpool organization, safety updates, and onboarding support. The design should be simple, intuitive, and community-focused.
Using a consistent prompt, design prototypes were created across Figma Make, Claude, and UX Pilot to enable comparative evaluation of outputs and workflows.
AI Platforms
Typography and Colors
AI tools including FontJoy AI Font Generator and Coolors Color Palette Generator were used to explore font pairings and adjust color palette variations, with accessibility standards guiding final selections.
FontJoy: Montserrat
Initial Palette
Coolors Modified Final Color Theme
Tone and Language Examples
Everyday Use
Find me a route for school.
Sure thing! The fastest route to Maple Middle School leaves at 7:15 a.m. from Pine Street. Want me to send you an alert 10 minutes before pickup?
Where is my ride now?
Your bus is on the way! It’s two stops away and should arrive in about 6 minutes.
I need to join a carpool.
Got it! I found two nearby carpool groups heading to your campus. Would you like me to show them?
Safety and Emergency Use
I feel unsafe walking home.
I’m here for you. I can share your location with your emergency contact or suggest a safer route. Which would you like me to do first?
Where is my ride now?
No worries—it happens! I can help you find another route or message your parent or driver. What would you like to do?
Help, I need emergency assistance!
I’m sending an alert to your emergency contact now. Stay where you are if it’s safe. I’ve also marked your location for responders
Design Implementation
Tone and Language
I have determined that setting a friendly, accessible, and calm tone will make SafeRide feel trustworthy for both students and parents, especially in safety or emergency contexts.
Pictured below is the prompt Input used to generate the sample UI text labels on the right (1) and tone and language guideline examples (2).
Prompt
I'm designing a SafeRide app for students, adult college student and parents and want the tone of the app to be friendly, accessible, and appropriate for emergencies. It should communicate clearly at a fifth-grade reading level. Can you generate sample chatbot responses or Ul text that reflect this tone? Provide other suggestions as well.
AI-enhanced Persona Defined
Meet Ava:
Ava is a 23-year-old graduate student who works as a part-time adjunct instructor at the university she attends.
User Scenario:
Ava recently moved to an apartment 2.5 miles off campus. While she still enjoys daytime walks on campus, her new location and busy schedule make it harder to do so regularly. In the evenings, Ava prefers using rideshare services and feels safer riding with fellow students.
Expectations:
-
Receive updates on available drivers quickly and easily
-
Feel confident knowing safety standards are always met.
-
Count on timely pick-ups and arrivals — because being reliable is part of being socially responsible.
Chatbot Design Narrative
User Flow Diagram
Competitive Research
Competitive research included a review of industry articles and company resources, including Medium and Uber. AI-assisted analysis was used to help identify best practices and actionable insights which informed feature and design recommendations for the SafeRide app.
AI Output Recommendations and Practices
-
In SafeRide, use data to suggest optimal bus routes or carpool matches automatically.
-
Display real-time tracking of buses or carpool vehicles, update estimated arrival times and trips. Notify users of deviations or delays.
-
Simplify and rationalize reporting options by use of color and visual hierarchy.
-
For safety-critical flows such as tracking route, confirming child arrival, raising alerts, streamline the process to 1–2 taps so users don’t get lost in menus.
Likes
Most participants are parents who seem to prefer “driving themselves” to their activities
It’s more “convenient” and feel they have “more control”
Dislikes
Some participants “dislike traffic”
Participants dislike “using their own gas”
It’s more “time consuming”
AI Solution
Participants found that an AI chatbot for the ride share app would “make life easier”
It would be “helpful for students because they won’t be late” and they can “save money on cabs”
Suggestions
Participants feel that the app would help them “feel that their child is safer”
Important features are to have “background checks of drivers”
Parents setting up their own rideshare group and “access to their child’s locations”
Key Insights
71.4%
Drive themselves or their children to extracurricular activities.
42.9%
Would be interested in using a chatbot to help find or organize a carpool group for themselves, a student, or child?
71.4%
Find it important to receive real-time updates (via chatbot) about a student's bus or carpool location?
6 out of 7
Find it helpful to have a chatbot that could answer questions about bus schedules and stop locations.
Prompting for Discovery Research
Research Strategy
I created the core qualitative survey questions and used AI-assisted support to expand the survey with targeted multiple-choice and open-ended items.
The goal was to evaluate the overall product concept and the SafeBot feature, validate key assumptions, and surface user priorities and unmet needs. The survey was distributed and analyzed using Google Forms, with seven participants contributing responses.
Output
ChatGPT (1) and IBM Consulting Advantage (ICA) (2), tools were used to support early concept development.
The output offered more thoughtful insights and user-friendly design direction for the SafeBot functionality.
Prompt Input
During early concept development, I explored how AI-supported ideation could accelerate branding and product positioning.
Guided by the project’s focus on safe, efficient transportation for students, I evaluated multiple naming directions and selected SafeRide for its direct, user-centered clarity.
The supporting tagline, “Smarter Routes, Safer Journeys,” reinforces both the functional and emotional value of the experience.
Prompting for Idiation
AI Tools
Below are the tools used to implement this project.
Qualitative Goals
Operational Efficiency
-
Enhance the daily transportation experience, Increase punctuality and overall efficiency for students and families
Safety & Trust
-
Reduce stress and safety concerns
for users
Engagement
-
Foster safe, communicative relationships within communities through the platform
User Growth Retention
Enable customizations real-time safety alerts and updates
Quantitive Goals
The ChatGPT platform was used to evaluate potential impacts based on industry standards. Starting with the problem statement and project description as the initial prompt, I gradually added additional requirements, building on each prompt to refine the outputs. This iterative process helped surface the quantitative goals outlined below.
Potential Impacts
Prompt
What are some reasonable quantitative business goals I can propose to clients, based on general industry standards and the information you currently have?
ChatGPT Reasonings
Prompt
Can you explain how you arrived at these metrics? What reasoning or data informed your suggestions?
User Growth Retention
5,000
active users within the first 6 months of launch.
Aligns with what schools or local communities can realistically achieve in a controlled rollout.
Engagement
70%
of users interacting with SafeBot at least once per week.
Metrics were chosen to reflect healthy engagement without app fatigue.
Operational Efficiency
15 - 20%
reduction in missed or late pickups.
The targets were set to reflect realistic efficiency gains achievable through automation.
Safety and Trust
100%
compliance with Data Protection standards and regulations
Align with industry norms and support a mission-driven focus on safety and transparency.
Target Audience
-
Parents who want reassurance and visibility into their child’s commute
-
Students grades 5+ who need simple and reliable ways to get to school or extracurricular activities
-
Adult college students looking for safe, efficient, and affordable transportation options
Overview
Duration:
August 18 - November 3 (12 weeks)
Role:
UX Design Lead
Problem:
Many families experience daily uncertainty and frustration due to unreliable school transportation systems. Inconsistent bus arrivals, lack of real-time communication, and limited accountability create anxiety for parents and disrupt students’ education.
Description:
SafeRide is a mobile and desktop application created to support the education and transportation sectors.
It provides users with safe and easy access to bus routes and stop information, while integrating with community rideshare groups.
Included as part of its concept is an AI component entitled SafeBot – an AI-powered virtual assistant designed to help users navigate transportation options in addition to simplifying and safeguarding daily transport operations.
What Worked Well
87.5% High Task Success Rate
A strong majority of participants were able to complete the primary task without assistance, indicating that the core user flow, navigation structure, and action hierarchy effectively support goal completion.
This suggests the interaction model is intuitive and aligned with user expectations.
AI Bot Request Flow Felt Intuitive Once Discovered
After locating the SafeBot feature, users were able to move through the request process with minimal confusion. The images indicates no drop-offs and low misclicks on the AI response screens.
This indicates that the conversational structure, prompts, and step-by-step guidance successfully reduce cognitive load during task completion.
Positive Sentiment Toward Visual Design and Color Palette
Participants responded favorably to the overall interface aesthetic, particularly the blue color palette.
Users associated the colors with feelings of trust, calmness, and safety — emotional attributes that strongly support the app’s purpose in the context of student transportation.
This reinforces that the visual design is not only appealing, but also aligned with the product’s emotional and functional goals.
Where Users Struggled
High Misclick Rate on the Home Screen
The image indicates that a significant number of participants selected incorrect elements when attempting to access the SafeBot feature.
This suggests that the feature’s placement, visual prominence, or labeling did not provide a strong enough visual cue, highlighting an opportunity to improve hierarchy and call-to-action clarity on the home screen.
Confusion Between Similarly Named Features
User feedback reported difficulty distinguishing between features with similar naming conventions (e.g., SafeBot, SafeRide, and other “Safe–” labels).
This indicates that the naming system may create cognitive friction and reduce scannability, signaling a need to refine labeling for clearer differentiation and faster recognition.
Limited Visibility of the SafeBot Feature
Several participants also suggested making the SafeBot feature (1) more prominent on the home screen.
This feedback reinforces the need to elevate its visual priority through placement, contrast, iconography, or persistent access patterns.
Logo Designs
I used multiple AI platforms, including Microsoft Copilot, to generate initial logo concepts. I then refined and developed the direction into a cohesive, visually compelling design aligned with the final color theme.
AI Generated Logo Variations
Final Design
AI Anaylsis of Insights
Conclusion of Analysis
In conclusion, the AI insights aligned with participant feedback, especially in identifying the home screen as a friction point. Among participants who responded, the sentiment scale average shows that 7 of the 8 participants rated the primary task easy to complete overall.
The high task success rate (100%) observed during testing suggests that the core user flow is intuitive once users are oriented within the interface.
While overall sentiment was positive, findings also highlighted opportunities to improve feature discoverability—particularly for the SafeBot entry point—which informed targeted design refinements for subsequent iterations.
Home Screen Iteration
Prompting for Discovery Research
Final Iterations and Output
Synthetic Data
Based on usability findings, the first iteration focused on improving SafeBot visibility through hierarchy and refined labeling, improving task clarity to reduce home screen misclicks.
mage (1) shows the initial screen, where primary options were not visually differentiated, creating decision friction. In image (2), clearer labeling and familiar iconography were introduced to strengthen hierarchy, reduce cognitive load, and support faster user decision-making.
Synthetic data was as developed to demonstrate how the app will manage data inputs and determine how AI will use the data to functionality support the app.
The Mural whiteboard workspace was used to organize ideas and address ethical requirements and identify datasets that would address functional needs.
Additional components such as Class Schedule and Schedule of Household Members were required to implement the AI-driven features properly which will help to understand how the data directly influences component generation and design.
A prompt was developed. Data was input into ChatGPT to generate the SafeRide synthetic data Excel document.
Prompt
Generate synthetic data for the following categories:
-
Date and Time
-
Scheduling
-
Availability
-
Geo Location
The "Scheduling" category includes the Class schedule, Activity schedule, Schedule of other household members and Bus schedule data.
The "Availability" category should include the availability of group members and the availability of group drivers.
The "Location" category should contain the home address and geo locations of Users.
User Journey