Overview
I completed a client zero project as part of the AI for UX/UI Designers course offered by ELVTR. The course explored how AI can be applied throughout the design process from research and ideation to prototyping, testing, and delivery.
My goal in taking this course was to expand my knowledge of AI tools and learn how to integrate them more effectively into my workflow. For me, this project wasn’t just about building an AI-powered chatbot. It was about experimenting with how AI can be used in out-of-the-box ways to enhance creativity, improve efficiency, and keep the user at the center of the process.
Problem Statement
Homeowners and renters often feel overwhelmed by lawn maintenance due to lack of knowledge, time, and expenses.
Homeowners and renters often feel overwhelmed by lawn maintenance due to lack of knowledge, time, and expenses.
Solution:
An AI-powered virtual assistant app that delivers personalized, step-by-step lawn care guidance to reduce stress and build user confidence.
An AI-powered virtual assistant app that delivers personalized, step-by-step lawn care guidance to reduce stress and build user confidence.
Deliverables
• UX audit
• Qualitative Study
• Executive Insights Brief
• AI-Generated Wireframes
• AI-Driven Style Guide
• Usability Testing
• Custom UI Components
• Capstone
UX Audit
The first step was to conduct a UX audit of an existing embedded AI assistant. I chose Lemonade’s claims assistant, Jim, which guides users through filing insurance claims by asking conversational questions, gathering details, and supporting documentation uploads.
I evaluated Jim using heuristic principles such as error prevention, visibility of system status, match between system and real-world language, and consistency. The analysis showed how Lemonade designed an assistant that makes a traditionally frustrating process feel simple and even humanized. Key strengths included its clean, familiar chat interface, proactive error handling through guided flows and quick-reply chips, and strong use of personalization (e.g., remembering a pet’s name). These heuristics demonstrated how Jim reduces user effort, limits opportunities for error, and increases trust by balancing automation with empathy. View the complete audit
I evaluated Jim using heuristic principles such as error prevention, visibility of system status, match between system and real-world language, and consistency. The analysis showed how Lemonade designed an assistant that makes a traditionally frustrating process feel simple and even humanized. Key strengths included its clean, familiar chat interface, proactive error handling through guided flows and quick-reply chips, and strong use of personalization (e.g., remembering a pet’s name). These heuristics demonstrated how Jim reduces user effort, limits opportunities for error, and increases trust by balancing automation with empathy. View the complete audit
Qualitative Research &
Executive Insights Deck
Executive Insights Deck
The next step was defining a research statement to clarify what I wanted to learn. Using ChatGPT, I created 3 proto-personas to ground the work in user needs. Based on my persona needs, I identified key questions to guide my qualitative research interviews.
I tested my interview questions out with a conversational ChatGPT test before finalizing an interview guide.
I conducted 2 user interviews for qualitative insights and a survey to my classmates to gain quantitive data.
I tested my interview questions out with a conversational ChatGPT test before finalizing an interview guide.
I conducted 2 user interviews for qualitative insights and a survey to my classmates to gain quantitive data.
AI Generated Wireframes & Style Guide
Generated using Figma Make
Color palette generated using Coolors.co & type scale generated using Peppercorn Figma plugin
User Testing
I created a low-fidelity prototype using Figma Make. I imported my test to Maze and conducted 5 usability tests to evaluate how effectively users could complete a core task. I asked the users to:"Imagine you are leaving town for a week and you want to pause your lawn care schedule. Using the chatbot, try to pause all tasks for 1 week."
While all participants completed the task, completion times varied. Long pauses suggested a need for clearer labels and confirmation feedback. One user wanted custom messaging beyond the preset options. View the working prototype
✨ Capstone: Bringing It All Together ✨
The capstone phase of this project focused on synthesizing research, usability findings, and AI-assisted exploration into a cohesive end-to-end product experience. My goal was to move beyond early experimentation and demonstrate how AI can support real user needs while maintaining strong UX principles, ethical awareness, and accessibility.
Interactive Prototype
From Chatbot to Intelligent Coach
Through research and testing, it became clear that users didn’t just want a chatbot that answered questions. They wanted a trusted guide that could help them build confidence over time. This insight shifted the direction of the product from a reactive assistant to a proactive lawn care coach.
The final concept evolved into a hybrid experience combining:
• Conversational AI for quick, contextual support
• Structured task planning and reminders
• Personalized recommendations based on location, lawn type, and user goals
• Decision support for DIY vs. hiring professionals
This positioned the assistant as both a learning tool and a decision-support system, helping users feel more capable rather than dependent on automation.
Designing With AI as a Creative Partner
AI played a meaningful role across ideation, research synthesis, wireframing, and UI generation. Tools like ChatGPT, Figma Make, and Google Stitch accelerated early-stage exploration and reduced blank-page friction. However, AI-generated outputs required human refinement. Many lacked hierarchy, accessibility nuance, or brand personality. I refined typography, strengthened color contrast, and elevated component structure using my own style guide. This project reinforced my belief that AI works best as a creative accelerator, not a replacement for design thinking.
Ethical, Inclusive, and Accessible by Design
Responsible AI use was foundational to this project.
• All research data was anonymized
• Recommendations were framed as supportive options, not prescriptive commands
• Color contrast and typography were tested against WCAG standards
• The chat experience was designed with clear states, feedback, and accessibility considerations
• The goal was to ensure the assistant builds trust, autonomy, and confidence without bias or manipulation.
Success Metrics
To evaluate effectiveness long term:
• Task completion rate
• User confidence and trust
• DIY vs. professional conversion
• Time to actionable resolution
• Engagement and retention
These metrics ensure the assistant delivers measurable value beyond novelty.
Final Reflections
AI transformed my workflow from linear to iterative and exploratory. It removed friction in early stages, allowing deeper focus on user needs and refinement. The most impactful insight: AI is most powerful when paired with human empathy, oversight, and craftsmanship. This capstone demonstrates how AI can be integrated thoughtfully across the design lifecycle to create experiences that are intelligent, ethical, and genuinely helpful.