01 · Client overview
Client overview
About Customer
The customer is an enterprise software provider delivering digital solutions across diverse industries, including Facility Management, Real Estate, and Food Services. The organization offers a corporate cafeteria management platform aimed at improving workplace dining experiences. The solution helps employees navigate challenges such as time constraints, dietary preferences, and inconsistent meal availability, which often result in suboptimal food choices. The company specializes in building customized, scalable software solutions that enhance operational efficiency and support business transformation.
02 · Challenge
Challenge
The Food Services Platform Provider faced customer experience challenges due to reliance on static FAQs and manual support processes, resulting in delayed responses, generic answers, increased user frustration, and higher operational overhead. A key technical challenge involved enabling the language model to access and reason over large volumes of frequently updated internal data while maintaining response accuracy and relevance.
The platform struggled to deliver personalized dining recommendations, where time constraints, dietary requirements, and lack of real-time menu updates often led to poor meal choices and user dissatisfaction. Additionally, the solution required a highly resilient, fault-tolerant, and secure infrastructure to handle fluctuating query volumes while meeting strict compliance and security standards.
03 · Solution
Solution
To fix the customer experience, Comprinno built an AI chatbot powered by Amazon Bedrock Agent and Amazon OpenSearch Service. This lets users talk to the app, get personalized food ideas, and place orders easily without browsing many screens, making the whole dining process faster.
Key Actions:
Conversational Ordering :
-
The chatbot's main job is letting users place orders and get the best food suggestions based on what they like, what they've ordered before, and any dietary needs.
-
It can also suggest cheaper or healthier options. Moreover, the chatbot helps users with placing orders.
Live Data Pipeline :
-
An Amazon EventBridge rule starts an AWS Lambda function every 3 minutes to constantly get the newest menu details from the Platform’s systems.
-
This keeps the system instantly updated on food availability, counter wait times, and daily specials, and also syncs the updated data with the OpenSearch database for periodic querying.
Smart Search :
-
The Lambda function creates special search tags (OpenSearch Indexes) from the live data
-
This enables Amazon OpenSearch Service to run smarter searches, making the food suggestions more relevant
Periodic Menu Recommendations :
-
When a user asks a question, a second AWS Lambda function, powered with the help of Action Group from Bedrock Agen,t finds only available dishes from OpenSearch.
-
It sends this menu information along with the user's question to the Amazon Bedrock AI model. The model then picks the 4-5 best dishes based on factors like taste and queue time, making sure the answer is always supported by current menu details
-
A second lambda function, again powered with the help of an Action group, lets us place orders by helping them to add the suggested items along with their quality to the cart.
Tracking :
-
All user questions and the suggested dishes are saved in Amazon DynamoDB to help analyze data and make future suggestions even better.
-
The platform continuously learns from user preferences, optimizing recommendations and enhancing overall dining experiences over time.
## Architecture Overview
This architecture demonstrates a secure, serverless, and highly available AWS infrastructure. It integrates security, fault tolerance, multi-AZ resiliency, and serverless compute for scalable, event-driven workloads.
- Networking Layer
A Virtual Private Cloud (VPC) spans three Availability Zones, providing network isolation and high availability.
-
Amazon CloudFront accelerates content delivery, and AWS WAF filters malicious traffic.
-
API Gateway securely exposes APIs to the compute layer.
-
Compute & Orchestration
AWS Lambda functions, with the help of Action Group,s handle core application logic in a fully managed, event-driven environment.
-
Serverless architecture provides automatic scaling and eliminates server management overhead.
-
Database Layer
Amazon DynamoDB provides a highly available, fault-tolerant NoSQL database distributed across multiple Availability Zones.
- Security & Compliance
AWS KMS encrypts sensitive data, IAM enforces access control, and CloudTrail tracks user and API activity for auditing.
- Monitoring & Logging
CloudWatch captures logs, metrics, and alarms for serverless functions and infrastructure components, ensuring operational visibility.
- Automation & Repeatability
Infrastructure is designed with fully managed services and multi-AZ deployment patterns, ensuring consistent, resilient, and repeatable operations.
04 · Results
Results
Benefits
-
Fast and Accurate Suggestions: Provides instant, reliable recommendations based on current data, helping users make quick choices.
-
Improved User Experience: Reduces effort and simplifies interactions, making the process smoother and more convenient.
-
Secure and Compliant: Uses AWS KMS and strict access controls to protect data and enforce access restrictions, helping meet privacy and security standards.
-
Scalable and Reliable: Serverless architecture and multi-AZ deployment ensure the system handles high traffic and remains available.
Conclusion
By leveraging Amazon Bedrock Agent and Amazon OpenSearch within a secure, serverless AWS architecture, the Food Services Platform was able to deliver a transformative user experience. The GenAI-powered chatbot provides timely, accurate, and personalized meal recommendations while ensuring high standards of security, compliance, and scalability. This solution not only enhanced employee satisfaction and reduced operational overhead but also established a resilient foundation for future innovation.