TL;DR Sloth: The Privacy-First Group Chat Summarizer

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady

What I Built

Waking up to 150 missed messages in a group chat is exhausting. Finding out if you actually need to do anything—or if people are just arguing over dinner plans—takes too much scrolling.

I built TL;DR Sloth to reward laziness. It is a web application that acts as your personal, highly efficient sloth assistant. Instead of reading the chaos yourself, you paste the exported chat log into the dashboard. The application instantly processes the conversation and hands you a clean, 3-part morning briefing:

1. The TL;DR: A one-sentence summary of the entire conversation.
2. Plans & Mentions: A bulleted list of actual plans made, or if anyone specifically asked you to do something.
3. The Drama: A quick summary of any arguments, tension, or funny moments so you're not out of the loop.

Because of TL;DR Sloth, anyone who hates scrolling gets to stay lazy without missing the important details.

Demo

• Live Application: https://tldr-sloth.vercel.app

(Feel free to try it out by pasting a messy group chat log!)

Code

The full project is open-source and available here:

Jayesh-JainX / TLDR-Sloth

Your lazy, privacy-first group chat summarizer. Let the sloth do the reading.

🦥 TL;DR Sloth

Your lazy, privacy-first group chat summarizer. Let the sloth do the reading.

---

📸 Demo & Live Application

Take it for a spin: https://tldr-sloth.vercel.app (Replace with your actual Vercel link if different)

(Note: Ensure your screenshot is saved as demo.png inside your frontend/public/ folder so this image loads perfectly on GitHub!)

---

💤 The Problem: Group Chats are Exhausting

You step away from your phone for two hours, and suddenly your group chat has 150 missed messages. People are arguing, plans are being made, and someone is asking you a direct question—but finding it means scrolling through endless memes and chaotic texting.

You could paste the chat into a generic AI like ChatGPT, but group chats contain your friends' personal lives, inside jokes, and private drama. Commercial AI tools store that data and use it to train their corporate models.

Nobody wants their private group chat drama harvested by…

View on GitHub

How I Built It

The application is built to be fast, reliable, and completely stateless:

• Frontend: Built with Next.js and Tailwind CSS, featuring a relaxing, earthy green UI to match the sloth theme.
• Backend: A decoupled Node.js & Express server running entirely in-memory with absolutely no database.
• Open-Source AI Core: The backend routes through OpenRouter to tap into the open-weight model meta-llama/llama-3.1-8b-instruct.
• The "Lazy" Engineering: Open-weight models can occasionally truncate JSON responses. Instead of building a complex retry mechanism, I engineered an aggressive Regex boundary-extraction algorithm that lazily grabs the required data even if the JSON brackets are cut off.

Why Does Open Innovation Matter?

Group chats contain highly sensitive, personal drama, inside jokes, and private plans.

If I built this tool using a closed commercial API like ChatGPT, I would be taking my friends' private lives and handing them over to a massive tech company to be logged and potentially used for model training. Nobody wants their personal group chat drama harvested by big tech.

Open innovation is the only way this tool can ethically exist. By utilizing open-weight models via a stateless proxy, I guarantee a strict "in-and-out" memory design. The chat log exists in the server's temporary RAM for exactly 4 seconds during inference. The second the summary appears on the screen, the data is instantly and permanently destroyed.

The Lazy Payoff

This weekend, my friends were planning a cabin trip. I stepped away from my phone for three hours and came back to 84 messages spanning three different chaotic topics.

Instead of reading the backlog, I pasted the log into the app. In exactly 4 seconds, TL;DR Sloth told me:

• TL;DR: The group finalized cabin plans, but Dave accidentally scheduled a work shift.
• Plans: Mike is driving and bringing the cooler. Alex is getting the keys.
• Drama: Sarah got annoyed at Dave for forgetting his schedule.

It saved me 10 minutes of scrolling, let me know I didn't need to bring a cooler, and allowed me to safely ignore the rest of the chat. Mission accomplished.