Meme-ify My Day

AI that roasts your chaotic schedule so you don't have to.

Comment

Inspiration

Let’s be honest—I am tired.

My life recently has been a blur of back-to-back meetings, deadlines, and a calendar that looks like a game of Tetris I am currently losing. I found myself getting too serious, over and over again. You know that feeling when you stare at your screen for 10 hours, your back hurts, and you realize you haven't drank water since breakfast? That is my baseline.

I realized that if I can't actually fix my chaotic schedule, the least I can do is laugh at it. I didn't need another productivity tool telling me to "optimize my workflow." I needed a digital friend to look at my screen and say, "Bro, go touch grass."

That was the spark for Meme-ify My Day. I wanted to build an AI that doesn't try to save me—it just roasts me. It’s my way of taking the seriousness of my packed life and turning it into pure, chaotic vibes.

What it does

Meme-ify My Day is an AI-powered "Roast Master" for your digital life.

Visual Judgment: You drag and drop a screenshot (your messy desktop, your packed calendar, or your broken code).

The Roast: The AI uses Llama 4 Vision to analyze the image and generates a savage insult, a "Burnout Score" (0-100%), and a meme caption.

Multimodal Experience: It doesn't just show text. It uses the browser's Text-to-Speech engine to read the roast out loud, triggers a "Vine Boom" sound effect, and displays a reactive GIF (like Elmo on Fire) based on your stress level.

"Fix My Life" Mode: A satirical feature that rewrites your schedule. It cancels your "Sync with Engineering" and replaces it with activities like "Scream into Pillow" or "Stare at Ceiling."

Share the Pain: It generates a downloadable meme card so you can post your failure on LinkedIn.

How we built it

Speed was the priority. I didn't have time for slow AI.

The Brain: We used Groq Cloud running the new Llama 4 Scout model. It’s the only model fast enough to analyze an image and return a JSON roast in under 2 seconds.

The Skeleton: Built with Next.js 15 and TypeScript for type safety.

The Look: We used Tailwind CSS to build a "Neo-Brutalist" aesthetic—bold borders, hard shadows, and high contrast. It matches the raw energy of the internet.

The Vibe: We used canvas-confetti for particles, html2canvas for generating the meme images, and the native Web Speech API for the voice.

Challenges we ran into

Building this wasn't smooth sailing.

The Model Ghost: Halfway through development, the Llama 3.2 vision model I was using got decommissioned by Groq. I woke up to errors and had to scramble to migrate to Llama 4 Scout. It was a panic moment, but the new model turned out to be even funnier.

The CSS Wars: I tried to be on the bleeding edge with Tailwind v4, but it clashed with the Next.js config. I had to perform a "hard downgrade" to v3 to get the neo-brutalist shadows to pop correctly.

The Audio Blockers: Modern browsers hate auto-playing audio. I wanted a dramatic sound effect to play immediately upon roasting. I had to architect a state management solution that tied the audio trigger specifically to the user's "drop" interaction to bypass the blockers.

Accomplishments that we're proud of

Sub-Second Judgment: Leveraging Groq's LPU means the roast hits you almost instantly.

The "Fix My Life" Algorithm: Teaching the AI to respond with valid JSON that includes a satirical schedule array was a prompt engineering challenge, but seeing it recommend "Nap" at 2 PM was worth it.

Full Multimodal Vibe: We successfully combined Vision, Text, Audio, and Animation into a single, cohesive experience without crashing the browser.

What we learned

I learned that sometimes, the best way to handle stress isn't to work harder—it's to code something stupid and fun. Building Meme-ify My Day reminded me why I love engineering: it’s not just about business logic; it’s about creating an experience that makes someone smile.

I also learned that Prompt Engineering is actually API Engineering. Getting an LLM to be "funny" while strictly adhering to a JSON schema for UI rendering is a delicate balancing act.

What's next for Meme-ify My Day

Battle Mode: Upload two calendars and let the AI decide whose life is more miserable.

Slack Integration: A bot that automatically roasts teammates who schedule meetings after 4 PM on Fridays.

Mobile App: So I can get roasted on the go.

next.js, typescript, tailwind-css, groq, llama-4, vercel, react, html2canvas, web-speech-api

Built With

Share this project:

Updates