AI Meeting Memory Assistant

Introduction:
Meetings are powerful — but remembering what was discussed is often a challenge. Important decisions, task assignments, and responsibilities often get lost in chats or notes.

To solve this, I built an AI Meeting Memory Agent that can:

Store meeting notesRemember past conversationsAnswer questions using memoryUse a fast LLM (Groq) for intelligent responses

This project demonstrates how AI agents with persistent memory can improve real-world productivity.

Problem statement:
In most teams:

Meeting notes are scatteredPeople repeatedly ask the same questionsTasks get forgottenContext is lost over time

We needed a system that doesn’t just chat — but remembers.

Solution:
I built an AI agent that combines:

🔹 Persistent Memory
Stores all meeting notes in a local JSON file.

🔹 Smart Retrieval
When a user asks a question, the system retrieves relevant past notes.

🔹 AI Reasoning (Groq LLM)
Uses Groq’s ultra-fast LLM to generate intelligent answers based on memory.

🔹 Streamlit UI
A simple, interactive web interface for users.

Tech Stack:

PythonStreamlitGrouq APIJSONdotenv

System Architecture:

User enters meeting note or questionSystem checks if input is a questionIf note → store in memoryIf question → fetch memory + send to GroqGroq generates contextual answerResponse is displayed in UI

Key Feature — Memory + Intelligence

Unlike traditional chatbots, this system:
✔ Remembers past inputs
✔ Uses context to answer questions
✔ Improves over time
✔ Acts like a real AI assistant

Example:

Input:
Shiv will attend the meeting
Sri is doing backend

Question:
Who is doing backend?

Output:
Sri is doing backend

sion

Why Groq?

I used Groq because:

Extremely fast inferenceFree tier availableSupports powerful open-source modelsIdeal for real-time AI agents

UI Preview

The interface includes:
Input box for notes/questions
Submit button
Memory viewer
Clear memory option

Challenges Faced:

API key integration issuesModel deprecation errorsIndentation bugs in PythonMemory retrieval tuning

Conclusion:
This project demonstrates how AI agents with memory can transform simple chat systems into intelligent assistants.

By combining:
Memory systems
LLM reasoning
Fast inference (Groq)

We can build practical tools for real-world productivity.