🧠 Meeting to Action Agent

📌 Overview

Meeting to Action Agent is an AI-powered system that transforms meeting discussions into structured action items and seamlessly syncs them with Google Calendar.

It analyzes meeting transcripts (entered as text or uploaded as .txt files), extracts tasks, assignees, and deadlines, and enables users to ask context-aware follow-up questions such as:

• What is Arjun’s task?
• Who has the nearest deadline?
• What did we discuss in the meeting?
• What is Riya’s deadline?

Built using FastAPI, Streamlit, LangChain, and Google Generative AI, this project bridges the gap between discussion and execution, turning meetings into actionable, trackable outcomes.

---

👨‍💻 Team Members

• N. Venkata Durga Karthik — Backend Development & AI Integration
• D. Naga Pallavi — Frontend Development (Streamlit UI)
• J. Sritha Reddy — API Integration & LLM-based Text Processing

---

⚙️ Tech Stack

Backend

• FastAPI
• Uvicorn

Frontend

• Streamlit

AI & Text Processing

• LangChain
• Google Generative AI
• Sentence Transformers

Calendar Integration

• Google Calendar API

Storage & Parsing

• FAISS
• Dateparser
• ics

Environment

• Local Development

---

🧾 Input Example

{
  "input_mode": "text",
  "meeting_transcript": "Riya will finalize the EcoGlow campaign design by Friday. Arjun will prepare the budget proposal by Wednesday.",
  "calendar_sync": true,
  "query": "Who has the nearest deadline?",
  "timezone": "Asia/Kolkata"
}

---

📤 Output Example

{
  "status": "success",
  "summary": "Meeting focused on finalizing the EcoGlow campaign and budget planning.",
  "generated_tasks": [
    {
      "assignee": "Arjun",
      "task": "Prepare the budget proposal",
      "deadline": "Wednesday"
    },
    {
      "assignee": "Riya",
      "task": "Finalize the EcoGlow campaign design",
      "deadline": "Friday"
    }
  ],
  "query_response": {
    "user_query": "Who has the nearest deadline?",
    "answer": "Arjun has the nearest deadline."
  }
}

---

⚙️ How to Run

1️⃣ Install Dependencies

pip install -r requirements.txt

---

2️⃣ Configure Environment Variables

Create a .env file in the root directory:

GEMINI_API_KEY = YOUR_GEMINI_API_KEY

⚠️ Do not commit your .env or credential files to GitHub

---

3️⃣ Run Backend (FastAPI)

uvicorn main:app --reload

---

4️⃣ Run Frontend (Streamlit)

streamlit run chat_app.py

---

🧩 How It Works

1. Users upload or paste meeting transcripts via the Streamlit interface.
2. The FastAPI backend processes the text using LLM-based analysis.
3. Tasks, assignees, and deadlines are extracted and structured.
4. Valid tasks are optionally synced to Google Calendar.
5. Users can ask follow-up questions to retrieve summaries, responsibilities, and deadlines.

---

🔐 Security Notes

• OAuth credentials and API keys are not included in the repository.
• Sensitive files are managed using .env and credentials.json.
• Only example configuration files are committed.

---

🏁 License

This is an open-source project, and contributions, experimentation, and learning are welcome.