Project Overview - Agent BHAVAN

This project contains the core logic for Agent BHAVAN, a multi-agent system designed to manage hotels. The agent is built using Google Agent Development Kit (ADK) and follows a modular architecture.

Problem Statement

Many hotels struggle with service gaps and operational inefficiencies that directly harm guest satisfaction and revenue. Guest support is often limited to frontline staff hours, leaving late-night or early-morning travelers without timely help and increasing frustration. Peak periods create long queues and slow check-ins/check-outs when manual processing is the norm, while inconsistent staff knowledge produces uneven information and service quality. Language barriers further shrink a hotel’s ability to serve international guests reliably.

Solution Statement

BHAVAN functions as a constantly accessible hospitality concierge that revolutionizes the way hotels engage with guests and oversee their operations. In contrast to systems limited by staff schedules or manual procedures BHAVAN offers round-the-clock assistance via natural chat communication, allowing guests to instantly reserve spa services, book dining tables, request late check-outs, or ask for additional towels without queuing or contacting the front desk. The agent provides precise and multilingual replies, eradicating human mistakes.

Architecture

BHAVAN is a modular, multi-agent hospitality system built on Google’s Agent Development Kit (ADK). A single root agent orchestrates a set of specialized sub-agents and function-tools so the system behaves as a composable ecosystem of focused capabilities (guest interactions, approvals, bookings, maintenance routing, upselling, analytics) rather than a monolith. This modularity improves testability, extensibility, and clarity of responsibility.

1. Latency reduction & availability :

To speed availability checks and room selection, BHAVAN uses a precomputed room map (JSON). Faster available room search, sink with database if any corruption occure.

2. Limitations :

• Local demo datastore only: Not connected to any real property management system; intended for development and testing.
• Mocked payments: Approvals and charges are simulated for demo purposes; do not represent real financial flows.

Essential Tools & Utilities

BHAVAN offers a targeted collection of utilities and agent-specific tools designed for room administration, confirmation handling and suggestions directed at guests. Every element fulfills a role, within the hospitality process.

1) Room Map utilities

A consolidated set of tools tasked with verifying room numbers setting up or refreshing the room map fetching available rooms and modifying room occupancy during check-in and check-out procedures. These features together uphold the hotel’s binary room availability map. Guarantee quick consistent room searches and alignment, with the demo datastore.

2) booking

Function-call utilities that modify room statuses in the datastore. These symbolize the operations supporting each booking and checkout processed by the system.

3) confirmation

Approval utilities that enforce a confirmation step before sensitive operations like booking or checkout are executed. These tools ensure explicit guest consent before finalizing any room-related action.

Technologies used

• Custom Tools
• Multi-Agent System
• Context engineering

Capabilities

• Prebooking
• Cancelling Prebooking
• Prebooking To Check-In
• On Time Booking (Check-In)
• Check-Out

File Structure

BHAVAN/
  ├── Agent/                             # Root agent and all ADK tool definitions
  │   ├── function_tools/                      # All function Tools
  │   │   ├── __init__.py
  │   │   ├── about.py                          # Function that returns hotel details
  │   │   ├── check_room.py                     # Function for room details checking
  │   │   ├── get_feedback.py                   # Function for storing feedback
  │   │   └── pay.py                            # Payment tool
  │   ├── sub_agents/                           # All subagents
  │   │   ├── sub_agent_tools/                  # Tools for sub agents                  
  │   │   │   ├── __init__.py
  │   │   │   ├── cancel_prebooking_tool.py     # Required functions to cancel prebooking
  │   │   │   ├── comfort_requests.py           # Required functions for comfort requests
  │   │   │   ├── prebooking_to_checkin.py      # Required functions check in from prebooking
  │   │   │   ├── prebooking_tool.py            # Required functions to prebook rooms
  │   │   │   ├── room_check_in.py              # Required functions for checking-in room
  │   │   │   └── room_check_out.py             # Required functions for checking-out from a room
  │   │   ├── __init__.py                       
  │   │   ├── booking_agent.py                  # Agent for booking room
  │   │   ├── general_agent.py                  # Agent for general comfort request
  │   │   ├── prebooking_agent.py               # Agent for prebooking room
  │   │   └── tourist_place_agent.py            # Agent for searching tourist spots (Not Implemented Now)
  │   ├── utils.py                              # All util functions
  │   │   ├── __init__.py
  │   │   ├── json_utils.py                     # Load / update / save room map (JSON)
  │   │   ├── room_map_utils.py                 # Room related functions
  │   │   └── utils.py                          # Util Configuration
  │   ├── __init____.py          
  │   ├── .env                                  # GOOGLE_API_KEY 
  │   ├── agent.py                              # Root Agent (BHAVAN)
  │   ├── config.py                             # Required configurations
  │   ├── database.py                           # Database connections
  │   ├── hotel_details.py                      # Hotel details
  │   ├── instruction.py                        # Instruction for Agents
  │   ├── logging.py                            # Logging file configuration (Not Implemented Yet)
  │   └── model.py                              # Database schema models
  ├── artifact/                                 # Artifact folder
  │   └── room_map.json                         # Room map
  ├── .gitignore                                # Gitignore 
  ├── hotel_management_db.db                    # Database that stores all datas
  ├── LICENSE                                   # License file 
  ├── README.md                                 # Readme file
  └── requirments.txt                           # Required libraries / modules

How To Run BHAVAN

1. Clone the repository

git clone https://github.com/ItsRikan/BHAVAN.git
cd BHAVAN

2. Create Virtual Environment

python3 -m venv venv          # For Linux/macOS
python -m venv venv           # For Windows

3. Activate the Virtual Environment

source venv/bin/activate      # For Linux/macOS
venv/Scripts/activate         # For Windows

4. Install dependencies

pip install -r requirements.txt

5. Create .env file inside Agent and write

GOOGLE_API_KEY = [your_google_api_key]

6. Go to BHAVAN directory and run the agent

adk web

7. Chose Agent from side bar dropdown