Case Study · 12 min read
73% of students feel "completely overwhelmed" by college admissions, yet most tools treat this $2.9B market like a search problem — not an emotional one. When Reach Best's CEO showed me screenshots of students crying in our feedback, I knew we had to redesign how AI talks to stressed teenagers. Four months later, our empathy-first platform was supporting 10,000+ students across 40 countries.
ROLE
Product Designer
TIME
8 Months
TEAM
Ryusei
CEO
Hritik
Full Stack Developer
Soufiane
Frontend Developer
Through 45 user interviews, one comment reshaped my perspective:
"Every app gives me the same list of schools, but none of them know that I cry myself to sleep worrying about my future."
Research
Insights
Archetypes
Prototyping
Testing
Launch
I mapped out the competitive landscape:
Naviance → Trusted but outdated; disengaging for students.
MaiaLearning → Widely used, but overwhelmingly counselor-focused.
Unifrog → Rich database, but complex navigation and poor student engagement.
From my research, I created student personas that reflected not just academic goals but also emotional
Riya Sharma
Discover universities that fit her strengths, interests, and ambitions.
Get clear guidance on essays, deadlines, and requirements.
Stay motivated while reducing stress and confusion.
Feels paralyzed by too many options and scattered information.
Struggles with anxiety about rejection and self-doubt.
Easily overwhelmed by excessive data and conflicting rankings.
Limited access to personalized counselor time.
Researches online late at night, mostly on mobile.
Creates “dream lists” of colleges but struggles to prioritize.
Relies on peer reviews and AI tools for instant feedback.
Motivated by progress tracking and milestone achievements.
Anil Verma
Help students efficiently discover and apply to best-fit universities.
Manage and track student applications in one place.
Provide data-driven, personalized guidance without burnout.
High student-to-counselor ratio leaves little time per student.
Overwhelmed by repetitive queries from hundreds of students.
Lacks easy-to-use tools to organize and visualize applications.
Prefers desktop tools for managing dashboards and data-heavy tasks.
Uses mobile notifications only for quick updates.
Relies on dashboards that give quick snapshots of student progress.
Values AI tools that reduce repetitive tasks and allow deeper mentoring.
These personas became the foundation for testing how empathy-driven AI could respond differently to different emotional patterns.
In testing with 12 students, 9 preferred personality-driven AI over generic responses. This validated that emotion could be a differentiator.
Test Setup:
A/B tested generic AI responses vs. personality-driven responses
Measured engagement time, return visits, and emotional feedback
Asked students to rate "helpfulness" and "understanding"
Results:
75% preference for personality-driven AI
3x longer engagement sessions
68% reported feeling "understood" vs. 12% for generic responses
Design Philosophy Evolution
My hardest design decision: Creating AI personalities that felt authentic without being manipulative.
The Challenge: How do you create AI that acknowledges emotional distress without exploiting vulnerability?
My solution: AI mentors that validated feelings first, then offered practical next steps—never dismissing emotions or rushing to solutions.
Initial Prototype Failure
My early prototypes came across as “chatbot therapy.” One student said:
"This feels like talking to a guidance counselor robot."
What Went Wrong
Over-analysis: AI was dissecting problems instead of acknowledging struggles
Clinical tone: Language felt therapeutic rather than peer-supportive
Solution-focused: Rushed to advice without emotional validation
The Redesign Process
I realized I was building AI that analyzed problems instead of AI that shared struggles.
Key Insight: Students didn't need therapy—they needed someone who understood their specific journey and could offer contextual guidance.
Design Pivot: shifting from clinical analysis to storytelling and emotional resonance.
Validation of New Direction
Engagement improved dramatically when AI mentors:
Validated feelings first ("This process is genuinely overwhelming—you're not alone in feeling this way")
Shared contextual understanding ("I remember feeling paralyzed by all the options too")
Offered practical next steps ("Let's break this down into one small step you can take today")
What I owned:
All user research and strategy development
Persona development and journey mapping
Complete UX/UI design process
Prototyping and usability testing iterations
Full design system architecture (built from scratch, no UI libraries)
Cross-platform design consistency (web and mobile)
Technical Constraints I Navigated:
AI Response Time: Balanced personality authenticity with sub-2-second response times
Scalability: Designed mentor personalities that could handle thousands of concurrent conversations
Mobile-First Reality: 67% of usage occurred on mobile devices, primarily late evening
Integration Complexity: Worked within existing school systems and data privacy requirements
🤖 AI Mentors: Personality-Driven Guidance
I created scalable story-driven personalities—dozens of them—like Maya the Transfer Student, Alex the STEM Explorer, and Jordan the International Applicant. These AI mentor personalities validated feelings first, then offered practical guidance.
AI Mentors - Chat Interface
Design Process:
Key Innovation:
Instead of generic advice, each mentor offered guidance through the lens of shared experience: "When I was navigating applications as a first-gen student, I felt the same way..."
Engagement Results:
✍️ AI Essay Review: Ownership Over Outsourcing
Feedback system that helped students refine their own essays instead of outsourcing them. (Research showed 84% wanted to "own their stories").
AI Essay
The Ethical Design Challenge:
Biggest influence moment: When stakeholders pushed for AI-written essays, I pushed back. Research showed 84% of students wanted to "own their stories." Instead, I proposed an essay feedback system that guided reflection without replacing student voices. This became one of our strongest differentiators.
How It Worked:
Impact: This ethical stance became a major conversion driver—parents and educators specifically chose us because we supported authentic student expression.
🔍 RB Search: Fit Over Rankings
Goal-based recommendations that curated schools based on fit, not rankings. AI explained why each school matched personal goals and preferences.
RB Search
The Algorithm Innovation:
Instead of ranking-based sorting, I designed a "Fit Score System" that considered:
Academic Profile Match: GPA, test scores, course rigor
Interest Alignment: Major preferences, career goals, research opportunities
Cultural Fit: Campus size, location, social environment preferences
Practical Considerations: Financial aid availability, distance from home
User Experience Design:
Instead of ranking-based sorting, I designed a "Fit Score System" that considered:
Explanation-First Results: Every recommendation came with clear reasoning
Interactive Filtering: Students could adjust preferences and see real-time changes
Comparison Tools: Side-by-side analysis of fit factors
Save & Organize: Personal lists with notes and deadlines
Technical Challenge: Balancing comprehensive data analysis with simplified decision-making interfaces—making complex algorithms feel intuitive and trustworthy.
📝 Smart Notes: Centralized Knowledge Management
A Medium-style editor for capturing insights, deadlines, and reflections—keeping admissions knowledge in one organized place.
Smart Notes
User Research Insight:
Students were losing critical information across multiple platforms—screenshots in photos, notes in various apps, deadlines on paper calendars. Additionally, they struggled to keep track of valuable resources they discovered while researching universities: helpful tweets from current students, YouTube campus tours, Reddit discussions about specific programs, and TikTok videos about campus life.
Design Solution:
Rich Text Editor: Formatting options for different types of content
Auto-Tagging: AI suggested relevant tags based on content
Deadline Integration: Calendar sync for important dates
Search & Filter: Quick retrieval of specific information
Sharing Options: Easy collaboration with counselors and parents
Behavioral Impact:
Students reported feeling 65% more organized and 43% less likely to miss deadlines after adopting Smart Notes.
🌍 Campus Exploration: Gamified Research
Integrated Google Earth with game-like controls, turning research into an interactive, tactile experience.
Campus View
The Innovation:
Traditional campus research felt overwhelming—static photos and endless text descriptions. I designed an immersive exploration experience that made discovering schools feel like a journey rather than a chore.
Features:
3D Campus Tours: Google Earth integration with custom waypoints
Interactive Hotspots: Click for specific building information, student reviews
Comparison Mode: Explore multiple campuses side-by-side
Personal Pins: Save favorite locations and add private notes
Weather & Seasons: See campuses in different conditions year-round
Engagement Boost:
Average session time increased from 4 minutes to 23 minutes when students used Campus Exploration vs. traditional search results.
📊 Progress Dashboard: Real-Time Student Visibility
Real-time visibility into each student's journey, including:
Which universities they applied to
Whether essays were submitted
If recommendation letters were uploaded
This gave counselors a clear picture of student progress without chasing updates over email.
Counselor Students View
Counselor Students View
Impact on Counselor Workflow:
73% reduction in administrative email exchanges
45% increase in time available for one-on-one student mentoring
89% of counselors reported feeling more connected to student progress
📈 Aggregate Insights: Data-Driven Decision Making
Trends across the entire student body, such as how many applied to a specific university or country. This helped counselors benchmark outcomes, allocate time, and prepare resources for the most popular destinations.
Counselor Analytics
Analytics Features:
Application Patterns: Most popular schools, programs, geographic preferences
Timeline Analysis: When students typically complete different milestones
Success Metrics: Acceptance rates, scholarship awards, enrollment decisions
Resource Planning: Predict workshop needs, counselor availability, deadline rushes
Strategic Value:
Counselors could proactively plan resources—if they saw 40% of students interested in engineering programs, they could schedule targeted workshops or bring in specific guest speakers.
🎯 Smart Filters: Focused Intervention
A quick way to surface disengaged or at-risk students, so counselors could focus their limited time where it mattered most.
Student levels
Risk Detection Algorithm:
Engagement Metrics: Login frequency, feature usage, response times
Progress Indicators: Missed deadlines, incomplete applications, stalled essays
Communication Patterns: Reduced interaction with AI mentors, declined help offers
Behavioral Changes: Sudden drops in platform usage or engagement quality
Counselor Interface:
Priority Queue: Students ranked by intervention urgency
Context Cards: Quick background on each student's situation
Suggested Actions: AI-recommended intervention strategies
Follow-up Tracking: Monitor effectiveness of outreach efforts
Result: Counselors could identify and support struggling students 3-4 weeks earlier than traditional methods, leading to better outcomes and reduced crisis interventions.
Multiple schools became paying customers within 6 months
Ethical positioning improved conversion rates significantly
Organic adoption reduced reliance on marketing spend
5M+ reach across TikTok and Instagram through organic user sharing
Invited to present at the Council of International Schools (only 2nd company after Duolingo)
Positive feedback from educators highlighted our ethical design stance
Featured in Mexican news channel and San Francisco Chronicle for our empathetic AI approach
73% reduction in administrative email exchanges
45% increase in one-on-one mentoring time availability
89% of counselors reported feeling more connected to student progress
Early intervention improved by 3-4 weeks on average
From Solo Designer to Systems Thinker and Story Architect
WHAT I LEARNED:
⚡
Empathy > Algorithms
⚡
Systems Thinking Through Constraints
With no UI kits or libraries, I was forced to think in systems and build a lean, scalable design language from the ground up. This constraint became a strength—every element was intentionally designed for our specific use case.
⚡
Ethical Design as Competitive Advantage
By choosing not to let AI write essays, we ended up building tools that supported deep thinking—and earned lasting trust. This ethical stance became our strongest differentiator.
⚡
Technical Constraints Breed Innovation
Working within school system limitations and mobile-first usage patterns forced creative solutions that became our most beloved features.
