Case Study — AI Tool

Writeup.ai

An AI-powered writing and speaking coach that gives IELTS, TOEFL, and Goethe candidates the one thing traditional tutoring couldn't: instant, consistent, affordable feedback.

RoleTeam Leader, AI-Driven Product Designer

Timeline2024

TeamHaidar Jbeily, Hasan Dayoub, Modar Ibrahim, Majd Sayouh, Kahtan Naser

ToolsFigma, FigJam, Maze, Illustrator, Jira, Microsoft Clarity

Learners using the platform

Cheaper than traditional tutoring

0×

Faster feedback turnaround

Personalized learning paths

Overview

Tutors couldn't scale. An algorithm could.

Writeup.ai is an AI writing and speaking assistant built around a personalized learning path. Every project I lead runs through the same five-phase process — this is how it played out here.

Discover

Understand the people and the problem before designing anything.

Define

Turn raw findings into a sharp problem statement and a testable hypothesis.

Ideate

Explore directions with HMWs, IA, and a value proposition before touching pixels.

Design

Build the flows, system, and UI that bring the concept to life.

Validate

Test with real users, measure against goals, and iterate until it works.

Problems we solved

Slow, subjective feedback

Traditional tutoring is slow, subjective, and costly, leaving learners frustrated.

Manual grading doesn't scale

Tutors face manual, time-consuming grading with no clear way to track progress.

No structured feedback loop

Learners need affordable, structured, and personalized feedback with clear improvement indicators.

Goals

Goal 1

Instant, rubric-aligned feedback

Deliver instant AI feedback aligned with official exam rubrics.

Goal 2

Visible progress

Enable progress tracking with clear improvement indicators.

Goal 3

Affordable, accessible prep

Provide affordable, accessible exam prep anytime, anywhere.

Discover

I led the interviews, surveys, and competitive audits myself — talking to both sides of the problem: the learners paying for feedback, and the tutors struggling to deliver it at scale.

Interviews

14 exam candidates

In-depth interviews with people preparing for IELTS, TOEFL, and similar language exams.

Interviews

5 tutors & institute managers

To understand the supply side — how feedback actually gets produced today, and where it breaks down.

Learner pain points

Students preparing for IELTS/TOEFL felt frustrated with slow, subjective, and costly tutor feedback, and struggled to track their improvement.

Tutor challenges

Educators said manual essay grading is time-consuming and inconsistent, making it hard to give data-driven feedback at scale.

Opportunity gap

Both sides wanted the same thing from different angles: instant, reliable AI feedback, progress-tracking dashboards, and personalized recommendations.

It's frustrating to wait days for feedback. I need to see my progress right away.Aya Ahmad, 24 — preparing for IELTS (B2C)

Manual grading takes too much time. I struggle to give consistent feedback to all my students.Sophia, 35 — IELTS tutor (B2B)

Interview and survey highlights — Interview & survey highlights

Define

The research pointed at one learner clearly enough to build a persona around her — and sharp enough to write a hypothesis I could actually test.

Aya Ahmad, 24

B2C — Persona 1

Goals

See progress right away, not days later
Understand exactly which skills are weak
Prepare without paying for a private tutor

Frustrations

Feedback is slow, subjective, and expensive
No clear way to track improvement over time
Generic advice that doesn't target her weak areas

"If I can track my improvement clearly, I'll feel more confident before the test." — Sallem Abdullah, 23, preparing for TOEFL

Problem Statement

Learners find manual tutor feedback slow, subjective, and expensive, with no clear way to track their progress over time — creating frustration and a lack of confidence.

Hypothesis

If Aya uses Writeup.ai's structured, personalized feedback system, she will feel more confident, identify weak areas faster, and improve her IELTS writing band score efficiently.

Ideate

Before any wireframe, I framed the problem as questions the whole team could ideate against — then checked our direction against a competitive audit.

HMWProvide instant, reliable feedback so learners don't have to wait days for tutors?
HMWTrack learner progress over time in a way that feels motivating and confidence-boosting?
HMWReduce the cost of exam preparation while maintaining high-quality guidance?
HMWSupport tutors in grading consistently and saving time without losing the personal touch?
HMWPersonalize feedback and study plans so each learner focuses on their unique weak areas?
HMWMake the preparation journey engaging so learners stay motivated until exam day?

Value proposition

Accessible anytime, anywhere Cost-effective vs. private tutors Reliable, rubric-aligned AI feedback Personalized to each learner's gaps Gamified for motivation

Design

From the HMWs, I mapped the information architecture and user flow, then moved into low-fidelity wireframes before building out a full design system.

Structure

Information architecture & user flow

Mapped the end-to-end path from onboarding to first AI feedback, shared early with stakeholders and engineering to scope effort and keep the launch on schedule.

System

Design system & UI kit

Built a component-based design system and UI kit for desktop and mobile, developed into interactive prototypes for accessibility and consistency.

Validate

Concepts were validated with usability and A/B testing across both sides of the market — B2C learners and B2B tutors — before anything shipped.

Method

Usability testing

Ran structured usability sessions with real exam candidates against the interactive prototype.

Method

A/B testing

Tested feedback presentation and onboarding variants with both B2C and B2B users to refine what shipped.

Reflection

What this project reinforced

Writeup.ai is where I first designed for two audiences with opposing needs inside one product — B2C learners who want speed and confidence, B2B tutors who want consistency and time back. The AI wasn't the product; the trust built around its feedback was.