/// Web App Case Study
RoastLine
A Journaling App that will Roast You
Category
Web App
Stack
Next.js, TAILWIND, SUPABASE, MORE...
Year
2025
01 / The Context
Built With
Live At
https://roastline.vercel.appA Journaling App that will Roast You
Current journaling apps suffer from high user drop-off rates because they lack active engagement. They act as passive storage rather than active dialogue. I saw an opportunity to disrupt this model.
RoastLine was conceptualized to test a core product hypothesis: could introducing friction, humor, and 'brutal honesty' via an AI agent actually increase daily active users (DAU) and 7-day retention?
To validate this, I led the end-to-end development of an MVP. I chose a bold Neo-Brutalist design system to signal a break from traditional 'zen' journaling apps, and built the backend on Supabase for rapid iteration.
02 / The Problem Space
Defining the core
user pain points.
01
User Retention Mechanics
Most journaling tools rely on intrinsic motivation, which is why I always abandoned them after a week. The product challenge was designing an extrinsic hook—the "roast"—that would bring a user back without feeling like a cheap gimmick.
02
AI Sentiment Accuracy
The core value proposition relied on the AI surfacing genuine behavioral patterns. Tuning the prompt to be funny but accurately reflective of the user's entries required extensive iteration and testing against my own journal data.
03
Performance & Flow State
Writing is a flow state. Any latency in saving or syncing data breaks the experience. The architecture had to guarantee rapid interactions, which is why I chose a Next.js + Supabase stack.
03 / Strategy & Execution
From hypothesis
to prototype.
1 Week
Conducted a competitive analysis of 5 popular journaling apps. Identified the core pain point: they act as passive storage rather than active dialogue.
2 Weeks
Mapped out the core user loop. Designed a Neo-brutalist UI in Figma to visually differentiate the product from "zen" competitors and reinforce the "brutally honest" brand positioning.
4 Weeks
Built the MVP using AI-assisted coding workflows. Integrated a local sentiment model to analyze entries and generate personalized, humorous insights.
1 Week
Dogfooding phase. I used the app daily to test the prompt tuning, ensuring the AI responses felt natural and actually motivated me to keep writing.
04 / Impact & Learnings
Measuring
success.
100%
Personal Retention
The friction actually kept me journaling.
<300ms
Interaction Latency
Optimized to maintain user flow state.
4 wks
Time to Market
From initial hypothesis to live MVP.