Memora: AI Meeting Memory Assistant

Memora is an AI-powered meeting memory assistant designed to summarize, track, and manage your team’s meeting knowledge. It acts like a virtual project manager that never forgets past meetings and always keeps your team aligned.

Key Features

• Upload meeting audio or transcript
• Automatic transcription + speaker diarization
• AI-powered meeting summaries
• Action item and decision extraction
• Long-term memory of projects & tasks
• Slack + email reminders of unfinished tasks
• Web dashboard for browsing meeting history and tasks

System Architecture

                       +-------------------+
                       |      Frontend     |
                       |  Next.js + Clerk  |
                       +-------------------+
                                 |
                                 v
                       +-------------------+
                       |      FastAPI      |
                       |  API Gateway +    |
                       |  LLM Task Router  |
                       +-------------------+
                                 |
         ------------------------------------------------
         |                      |                      |
         v                      v                      v
+----------------+    +-------------------+   +------------------+
| Speech-to-Text |    |   LLM Processor   |   | Notification Svc |
| (Whisper /     |    | (Summary + Tasks) |   | (Slack + Email)  |
| AssemblyAI)    |    +-------------------+   +------------------+
+----------------+
         |
         v
+--------------------------------------------------+
|               PostgreSQL + Weaviate              |
| Project Memory: Meetings, Tasks, Summaries, Vectors |
+--------------------------------------------------+

Data flow:
Frontend → API → Process (STT + LLM) → Store (Postgres + Vector DB) → Notify User

Tech Stack

• Backend: Python, FastAPI
• Speech-to-Text: OpenAI Whisper / AssemblyAI
• LLM: LLaMA 3, GPT-4o, or Claude
• Database: PostgreSQL + Weaviate (vector memory)
• Frontend: Next.js, TypeScript, shadcn/ui
• State Management: Zustand
• Authentication: Clerk
• Emails: Resend
• Hosting: Vercel + AWS

Folder Structure

memora/
├── backend/
├── frontend/
├── models/
├── data/
├── docs/
├── scripts/
├── docker/
├── .env.example
├── README.md
└── LICENSE

---

How to Run Locally

1. Clone the repo

git clone https://github.com/yourusername/memora.git
cd memora

2. Setup backend

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
uvicorn app.main:app --reload

3. Setup frontend

cd frontend
pnpm install
pnpm dev

OR

cd frontend
npm install 
npm run dev

MVP Feature Backlog

This is the suggested feature task list for first release.

Priority	Feature	Notes
High	Audio Upload System	User uploads .mp3, .wav
High	Speech-to-Text Pipeline	Whisper or AssemblyAI
High	Speaker Diarization	Separate "who said what"
High	LLM Meeting Summary	GPT-4o / LLaMA 3 basic summary
High	Task & Decision Extraction	Identify owners + deadlines
Medium	Store Sessions to DB	PostgreSQL schema for projects
Medium	Slack / Email Reminders	Unfinished tasks reminders
Medium	Web Dashboard	Next.js UI for meetings + tasks
Medium	Authentication	Clerk integration
Low	Google Calendar Integration	Auto fetch participants + times
Low	Real-time Note Taking Agent	(Advanced demo only)

Roadmap

• Audio upload + STT pipeline
• LLM-powered meeting summary
• Task + decision extractio
• Long-term memory DB
• Slack/email reminders
• Web dashboard
• Google Calendar integration