AI Operations Assistant

A local AI operations assistant that processes natural language tasks using a multi-agent architecture. The system fetches real-time weather and news data, then generates intelligent summaries.

Architecture

The system uses three specialized agents working in sequence:

Planner Agent: Analyzes the user's natural language task and creates a structured execution plan
Executor Agent: Executes the plan by calling external APIs (weather, news)
Verifier Agent: Validates results and formats a comprehensive response with AI-generated summary

APIs Used

Open-Meteo API: Free weather data with no API key required
- Geocoding for city coordinates
- Current weather conditions
Google News RSS Feed: Free news headlines with no API key required
- Location-based news search
- Top headlines

Prerequisites

Python 3.8 or higher
Google Gemini API key (free tier available)

Setup

1. Clone or Download

Ensure all project files are in the ai_ops_assistant directory.

2. Install Dependencies

pip install -r requirements.txt

3. Configure Environment Variables

Copy the example environment file:

cp .env.example .env

Edit .env and add your Gemini API key:

GEMINI_API_KEY=AIzaSyC1n_D1Ahmq9l3T5fk9WF0fJYl5uY2gIUs

just copy this to your newly made .env file

Running the Application

Start the FastAPI server:

uvicorn main:app

The server will start at http://localhost:8000

Usage

Send a POST request to /process with a natural language task:

curl -X POST http://localhost:8000/process \
  -H "Content-Type: application/json" \
  -d '{"task": "Give today'\''s weather and top news for Delhi"}'

Example Prompts

Weather and News for a City:

{"task": "Give today's weather and top news for Delhi"}

Mumbai Weather and Headlines:

{"task": "What's the weather like in Mumbai and what are the top news stories?"}

Bangalore Information:

{"task": "Get current weather and latest news for Bangalore"}

Chennai Updates:

{"task": "Show me Chennai weather and today's headlines"}

Kolkata Overview:

{"task": "Provide Kolkata weather forecast and top news"}

Response Format

The system returns a structured JSON response:

{
  "city": "Delhi",
  "weather": "Partly cloudy, 24°C, Humidity: 65%, Wind: 12 km/h",
  "top_headlines": [
    {
      "title": "News headline 1",
      "source": "Source Name",
      "link": "https://..."
    }
  ],
  "summary": "AI-generated summary of the weather and news...",
  "timestamp": "2026-02-04T21:25:00.123456"
}

Project Structure

ai_ops_assistant/
├── agents/
│   ├── planner.py      # Task planning agent
│   ├── executor.py     # Tool execution agent
│   └── verifier.py     # Response verification agent
├── tools/
│   ├── weather.py      # Open-Meteo weather integration
│   └── news.py         # Google News RSS feed parser
├── llm/
│   └── client.py       # Gemini LLM client with JSON output
├── main.py             # FastAPI application entry point
├── requirements.txt    # Python dependencies
├── .env.example        # Environment template
└── README.md           # This file

How It Works

User sends a task → FastAPI receives the request
Planner analyzes → Identifies required tools (weather/news) and extracts parameters (city name)
Executor fetches data → Calls Open-Meteo API for weather and Google News RSS for headlines
Verifier formats → Validates data, generates AI summary, returns structured response

Known Limitations

Requires valid Gemini API key for operation
Weather data limited to current conditions (no forecasts)
News limited to publicly available RSS feeds
City names must be recognizable by the geocoding service
Response time depends on external API availability
Summary quality depends on LLM performance

Error Handling

The system gracefully handles:

Invalid city names → Returns error in weather field
API failures → Captures errors without crashing
Missing data → Provides partial results when possible
Network timeouts → Reports timeout errors

License

This is a demonstration project for educational purposes.

Ai Ops Assistant