🌦️ Weather Agentic AI

An intelligent, conversational weather assistant built with Langchain, Streamlit, Firebase, and ElevenLabs. Ask weather-related questions via text or voice, get precise answers, and hear them spoken back to you.

---

🧠 Features

• 💬 Conversational weather assistant
• 🎙️ Voice input using ElevenLabs Speech-to-Text
• 🔊 Voice responses with ElevenLabs Text-to-Speech
• 💾 Chat memory with Firebase (multiple chat sessions)
• 🧭 Real-time weather data fetching via LLM agent
• 🧹 Delete individual chat sessions
• 📋 Intuitive chat-style UI using Streamlit

---

📸 Demo

---

🚀 Quick Start

1. Clone the Repo

git clone https://github.com/Saon110/Weather_Agentic_Ai.git
cd Weather_Agentic_Ai

✅ 2. Create a Virtual Environment

# For Linux/macOS
python3 -m venv env_name
source env_name/bin/activate

# For Windows
python -m venv env_name
env_name\Scripts\activate

3. Install Requirements

pip install -r requirements.txt

requirements.txt includes:

• streamlit
• speechrecognition
• pyaudio
• elevenlabs
• firebase-admin
• requests
• python-dotenv

3. Setup API Keys

Create a .env file with the following:

GROQ_API_KEY=groq_api_key
OPENWEATHERMAP_API_KEY=openweathermap_api_key
ELEVENLABS_API_KEY=elevenlabs_api_key
ELEVENLABS_VOICE_ID=21m00Tcm4TlvDq8ikWAM
ELEVENLABS_MODEL_ID=eleven_monolingual_v1
FIREBASE_TYPE=service_account
FIREBASE_PROJECT_ID=weather-ai-agent-1c82f
FIREBASE_PRIVATE_KEY_ID=firebase_private_key_id
FIREBASE_PRIVATE_KEY=firebase_private_key
FIREBASE_CLIENT_EMAIL=client_email
FIREBASE_CLIENT_ID=client_id
FIREBASE_CLIENT_X509_CERT_URL=client_certificate_url #this informations can be downloaded upon creating a firebase database

4. Run the App

streamlit run app.py

---

🗂 Project Structure

.
├── app.py                  # Main Streamlit app
├── firebase_chat.py        # Firebase chat handling
├── firebase_setup.py       # Firebase setup
├── weather_agent.py        # Weather reasoning agent
├── voice_utils.py             # Combined voice input/output with ElevenLabs
├── requirements.txt
└── .env

---

🛠️ TODO

• Add unit tests
• Add a frontend for selecting city/location
• Deploy on Streamlit Cloud or Hugging Face Spaces
• Support more languages in STT/TTS

---

🧑‍💻 Author

Made by Sijon Chisty Saon

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.