Bedrock Agent for Hotel Booking - AI Assistant with Tools, Action Groups & RAG with Knowledge Bases

This project simulates how hotel chains can offer smarter, AI-powered booking experiences using Amazon Bedrock Agents.

Built with Claude, Knowledge Bases, and Lambda-based Action Groups, this assistant can:

• Respond to queries about different types of rooms
• Check room availability for specific dates
• Book the selected room for the user

This mirrors how real-world hospitality companies are integrating conversational AI to reduce support load and improve bookings, with backend automation and retrieval-augmented generation (RAG).

What the Agent Can Do

User Request	Tool/Action Used	Backend Resource
"What types of hotel rooms do you have?"	Bedrock Knowledge Base (RAG)	Hotel details PDF in S3
"Is the sea-facing room available this weekend?"	Action Group 1 (Lambda)	Room Inventory Table in DynamoDB
"Book it for me"	Action Group 2 (Lambda)	Booking Table in DynamoDB

This Bedrock Agent uses a combination of LLM reasoning, RAG, and real-time actions:

Components Description

1. Bedrock Knowledge Base (RAG)

• Purpose: Handles general inquiries about hotel room types
• Data Source: Hotel details stored as PDF in Amazon S3
• Technology: Uses Retrieval-Augmented Generation (RAG) model

2. Action Group 1 (Lambda)

• Purpose: Checks room availability
• Data Source: Room Inventory Table in DynamoDB
• Functionality: Verifies availability of specific room types for requested dates

3. Action Group 2 (Lambda)

• Purpose: Processes booking requests
• Data Source: Booking Table in DynamoDB
• Functionality: Creates and manages booking records

4. Data Storage

• Amazon S3: Stores static hotel information documents (PDF format)
• DynamoDB:
  • Room Inventory Table: Tracks available rooms and their attributes
  • Booking Table: Maintains reservation records

What’s Provisioned via AWS CDK

Resource	Purpose
2 DynamoDB Tables	One for availability lookup, one for confirmed bookings
2 Lambda Functions	Used as Bedrock Agent tools (defined via OpenAPI)
S3 Bucket	Stores hotel room descriptions (PDF format) used for retrieval
AWS IAM Roles & Policies	Give Lambdas access to DynamoDB, S3

How the Rest Was Set Up

Everything Bedrock-related agent creation, knowledge base setup, tool registration, and OpenAPI integration was done manually via the AWS Console. You’ll find a full walkthrough of that process in this article

Deploying the Infra

git clone https://github.com/mohsinsheikhani/bedrock-hotel-booking-agent.git
cd bedrock-hotel-booking-agent
npm install
cdk deploy

Once deployed:

• Upload your hotel PDF into the hotel-assets bucket
• And follow the article: Hands-On with Amazon Bedrock Agents: Hotel Booking Assistant with Action Groups and Knowledge Bases

Demo

🙋‍♂️ Contact

Created by Mohsin Sheikhani
From Code to Cloud | Hands-on Cloud Architect | AWS Community Builder | Serverless | Infrastructure as Code (IaC) | Systems Design | Event-Driven Designs | GenAI | Agentic AI | Bedrock Agents | 3x AWS Certified

🚀 Follow me on LinkedIn for more AWS content!