ChatSumAI

This web app can summarize your tons of messages and get some greate and valuable insights and divide those mess messages into understandable and sorts them importance wise.

Comment

chatsumAI — "What Did I Miss?"

License: MIT Built with: React + Vite + TypeScript Local-First

Catch up without reading everything.
MissedIt AI extracts key decisions, tasks, personal mentions, unanswered questions, and deadlines from messy conversation transcripts — presented in a prioritized dark-mode dashboard with verifiable source evidence.

💡 The Core Problem

Users often return to hundreds of unread messages across WhatsApp, Slack, Discord, or Linear without the time to read through line-by-line. Critical deadlines get buried, personal mentions get missed, and decisions get lost in casual chatter.

MissedIt AI answers three questions immediately:

  1. What happened? (Concise overview & participants)
  2. What matters to me? (Direct asks & personal mentions)
  3. What should I do next? (Interactive action checklist & deadlines)

🔒 100% On-Device Privacy Guarantee

Built to comply with strict individual hackathon privacy and zero-data-retention requirements:

  • Zero Remote AI APIs: No OpenAI, Anthropic, or cloud LLM endpoints receive conversation text.
  • Zero Backend Database: Operates entirely client-side in the browser.
  • Zero Persistent Storage: Does not store conversation transcripts or results in localStorage, sessionStorage, cookies, or IndexedDB.
  • Ephemeral RAM Execution: Conversations exist exclusively in temporary browser memory and are completely wiped when you click "Clear Conversation" or refresh the tab.
  • Zero Network Transmission: Verified with client-side network audits.

✨ Key Features

  • Multi-Format Chat Parser: Ingests exports from WhatsApp ([10:14 AM] Jordan: ... and 10/9/26, 10:14 AM - Jordan: ...), Slack/Discord (Jordan [10:14 AM]: ...), and plain text threads.
  • Deterministic Heuristic Engine:
    • Detects explicit deadlines & dates (before 1:00 PM today, by Thursday 8:00 PM, today, tomorrow).
    • Identifies personal assignments & @mentions based on your customizable name & aliases.
    • Distinguishes completed tasks (marked completed, finished and pushed) from pending commitments.
    • Filters out casual chatter (pizza arrived, grabbing coffee, sounds good) without false alarms.
    • Detects team decisions & architecture agreements (let's stick with SQLite, locked sequence).
    • Surfaces unresolved questions and follow-ups.
  • WhatsApp Authentication & Secure Onboarding:
    • Phone Registration: Sign up using your WhatsApp-registered phone number with international country code selection.
    • WhatsApp OTP Verification: Receive and verify a 6-digit OTP code (with simulated instant WhatsApp notification toast and one-tap auto-fill).
    • Password Security: Create an account password with strict validation requiring at least 8 characters and real-time security checks.
    • Sign In / Log In: Existing users can log in with their WhatsApp phone number and password or instant one-time passcode.
  • Interactive Action Checklist: Check off tasks during your triage session with visual strikeout and completion state.
  • Verbatim Evidence Citations: Every extracted task, decision, and question links directly to the original quote, sender, and timestamp.
  • Local Digest Export: Download intelligence briefings in formatted Markdown (.md) or raw JSON (.json) generated 100% in-browser via native Blobs.
  • Interactive Edge Case States: Dedicated preview tabs for Analyzing/Loading, Helpful Empty Dropzone, Friendly Error Validation, and All Clear (Inbox Zen).

🚀 Getting Started

Prerequisites

  • Node.js (v18+ or v20+ recommended)
  • npm

Installation & Local Run

# Clone the repository
git clone https://github.com/your-username/misseditai.git
cd misseditai

# Install dependencies
npm install

# Start development server
npm run dev

Visit http://localhost:5173 to test the application locally. web link : https://mubarakaist-sudo.github.io/chatsumai/

Production Build

# Compile TypeScript & build static bundle
npm run build

# Preview static distribution
npm run preview

The optimized static build is emitted to dist/ and is ready for static deployment on GitHub Pages, Vercel, or Netlify.

🧪 Acceptance Criteria & Verification

All requirements from Section 10 of the Product Requirements Document (PRD) are implemented and verified:

Requirement Status Implementation Details
No Backend Required ✅ Pass 100% client-side React + TypeScript application
User Input & Paste ✅ Pass Large textarea with character meter, paste button & clear
Sample Conversation ✅ Pass Pre-loaded CS Capstone sprint with deadlines & assignments
Deadlines & Decisions ✅ Pass Detected before 1:00 PM today, SQLite decision & sequence
Source Evidence ✅ Pass Verbatim quotes with collapsible view on all findings
Casual Chat Filtering ✅ Pass Pizza & lunch chatter excluded from task extraction
Error Handling ✅ Pass Friendly guidance for empty or < 2 message input
Checklist & Reset ✅ Pass Toggleable checklist items & memory-wiping Clear button
Zero Data Leakage ✅ Pass In-memory RAM execution only, no network calls

Run the automated test suite:

node --experimental-strip-types test-verification.mjs

🎨 Design System

Styled according to the Stitch design specifications:

  • Canvas Base: #0E141C (Deep void obsidian)
  • Surfaces: #161C24 (Layer 1) & #1A2028 (Layer 2)
  • Accents: Primary Mint (#10B981 / #4EDEA3), Violet (#8B5CF6), Coral Urgent (#F43F5E)
  • Typography: Plus Jakarta Sans (Headlines) & Inter (Analytical body/metadata)

📄 License

MIT License. Built for individual hackathon evaluation.

Built With

Share this project:

Updates

Submission history