Back to work

Overview — Education with AI

Overview: education was breaking under its own weight. Teachers were drowning in paperwork, students were passing through a system that never stopped to ask if it was actually working for them. This product is an AI-powered SaaS platform connecting teachers and students, automating the most time-consuming parts of teaching while giving every student a learning experience built around who they are. The platform was not built to replace teachers — it was built to give them back the time, energy, and focus quietly stolen by administrative work.

What I did

What I did: Requirement Analysis facilitating stakeholder sessions, User Research interviewing and observing teachers and students, Competitor Benchmarking of the ed-tech landscape, Design Execution across user flows, IA, wireframes and high-fidelity prototypes in Figma, Cross-Functional Collaboration with developers and the client, and Interaction Design & Micro Animations.

The weight teachers carry

The weight teachers carry: teachers spend nearly 50% of their time on non-teaching tasks, 72% report burnout, and students sit in classrooms designed for the average learner, not for them.

The problem

Problem 1, administrative overload: creating a single exam or assignment manually takes 2-3+ days, 50% of time goes to non-teaching tasks, 72% of educators report burnout. Problem 2, generic learning models: one curriculum delivered the same way to every student, with no visibility into individual struggles, no tailored pathway, and no motivation for students who can't see their progress. Problem 3, high operational costs and no unified system: personalized education requires more staff and more cost, while tools stay siloed across an LMS, a grading tool, and spreadsheets.

The gap, objective and success criteria

The gap in the market: no single platform unified grading, content generation, and personalized learning into one connected, AI-first ecosystem teachers could trust. Objective: design an AI-powered MVP that automates grading and content creation, enables personalized learning paths, and reduces operational overhead — validated with 10 pilot institutions before scaling. Success criteria table covering reduced teacher workload, improved grading efficiency, personalized learning, teacher trust, institutional fit, and AI accuracy.

Research

Research: early sessions aligned on institutional goals, technical constraints, and the core product vision, with one principle held throughout — AI should assist teachers, not replace them. User research findings from teachers and students, plus a table mapping each finding to its design response, such as side-by-side AI review with human override, AI content generation with edit controls, workbook digitization, gamification, and an admin analytics dashboard.

Three people, one platform — our process

Three people, one platform: personas for Amara the teacher, Kavin the student, and Nilufar the system administrator, each with their own goals and frustrations. Our process: a repeatable Discover, Define, Design, Review, Iterate loop used to design, verify, and ship one screen at a time.

Mapping the journey

Mapping the journey: three portals, three distinct navigation systems, one shared AI engine underneath. The Teacher portal covers dashboard, content generation, automated grading, gradebook, and video conferencing. The Student portal covers dashboard, assignment submission, personalized learning, and video conferencing. The Admin dashboard covers system overview, user management, and content management.

Solution & key features

Solution and key features overview, followed by final UI screens for the Teacher flow (content generation and assigning), the Student's flow (personalized learning), and the Admin flow (dashboard).

Did it work?

Did it work? The MVP was tested with 10 pilot institutions: 80% reduction in content prep time, 10-24 hours saved per teacher per week, 92% AI grading accuracy, and a 30% boost in student engagement, with impact broken down for teachers and for students.

What we learned

What we learned: every feature that resonated came from observed behaviour, not brainstorming; committing to the 'AI assists, never replaces' principle built teacher trust; and badges/progress bars tied to real performance drove motivation. What needs improvement: visual consistency across the three portals needs tightening, and feedback states such as loading, empty, error and success need to be designed intentionally.
Back to all projects