Project Overview

Barmetrix, a leader in bar and restaurant inventory management, sought to speed up its reporting process by generating automated email templates and inventory charts using Generative AI. Avahi proposed a project solution on AWS featuring Amazon Bedrock, AWS Lambda, and other services to streamline the data flow and reduce the time needed to deliver inventory reports. The result was a three-week engagement fully funded by AWS, yielding a working project that demonstrates fast, reliable, and automated report generation capabilities.

About the Customer

Barmetrix is a bar and restaurant inventory management specialist operating across North America. The company’s services focus on helping hospitality businesses improve profitability and efficiency through meticulous inventory control and actionable reporting.

The Problem

With traditional processes, Barmetrix faced delays creating and customizing bar inventory reports for its clientele. These manual tasks involved gathering data from a Microsoft Azure SQL Database and turning them into clear, concise reports. Failure to reduce turnaround times could have led to inefficiencies, slower decision-making for Barmetrix’s customers, and missed opportunities to scale the business on a larger, more automated foundation.

Why AWS

AWS offered Barmetrix a wide range of AI-driven services, such as Amazon Bedrock and AWS SageMaker, that could be quickly implemented and tested in a project environment. The scalability and reliability of AWS allowed Barmetrix to trial new AI functionalities within a three-week window, without committing to costly physical infrastructure.

Why Barmetrix Chose Avahi

Avahi stood out due to its proven track record with Generative AI solutions and advanced AWS partnerships. The Barmetrix team needed a partner that understood both the hospitality industry’s
operational pressures and the technical complexities of AI/ML on AWS. Avahi’s clear planning, promptengineering experience, and ability to leverage AWS funding made it an ideal choice to execute this project quickly and effectively.

Solution

Avahi implemented a middleware layer that connects Barmetrix s Azure SQL data to AWS-based AI services. The project uses Amazon Bedrock for generative model inference and AWS SageMaker for potential fine-tuning or future productionizing of ML workloads. A basic React UI was developed to demonstrate how Barmetrix s end users could interact with:

Project Impact

By deploying the solution on AWS, Attorney Live now offers an enhanced, near-instant response system to users seeking immediate legal guidance. The integration of intelligent natural language processing and robust data indexing has improved user satisfaction and streamlined internal support processes.

• Automated Email Templates: By integrating Claude 3.5 prompt engineering into an AWS Bedrock workflow, Barmetrix can generate tailored emails for bar operators based on real-time inventory data.
• Inventory Charts and Recommendations: Leveraging AWS services to ingest data from S3, the system produces up to 10 visual charts of key items and provides data-driven recommendations. This guidance helps bar owners quickly see overstock or understock trends.
• API Interactions: Amazon API Gateway front-ends the service, enabling secure, scalable requests. AWS Lambda handles triggers for chart generation and template creation.
• Data Storage: Pinecone VectorDB further supports embeddings storage for advanced data retrieval, while the entire project is protected and monitored using standard AWS best practices.

 Key Deliverables

• Basic React UI for automated email template generation
• AI model integration via Claude 3.5 and AWS Bedrock
• Automated charting and inventory recommendations
• Middleware hosted on AWS to connect Azure SQL data
• End-to-end testing and a minimal documentation handoff

Project Impact

By completing this project, Barmetrix successfully demonstrated the feasibility of rapid, AI-driven report generation. They now have a clear roadmap for scaling these capabilities in the future.

• Completed project in 3 weeks timeframe
• Project fully funded by AWS $25,000 in AWS credits
• Managed feedback and validation within a 3-day turnaround