Build a Retail Copilot Code-First on Azure AI

This repository is a companion to the WRK550 workshop session offered on the 2024-2025 iteration of the Microsoft AI Tour, a worldwide tour of events. Register to attend an event near you!

Session Description

In this workshop, attendees will be introduced to the concept of GenAIOps, and learn to build a custom RAG-based retail copilot end-to-end, code-first on the Azure AI platform.

Using the provided sample, attendees will learn to prototype, evaluate and deploy a customer support chat AI for Contoso Outdoors - a fictitious retailer who sells hiking and camping equipment online. The sample uses a Retrieval Augmented Generation (RAG) architecture to implement a retail copilot solution that responds to customer queries with answers grounded in the retailer's product catalog and customer purchase history.

Application Architecture

Attendees will also learn the components of an Azure AI application architecture for implementing a custom copilot solution. User input ("question") to a frontend chat UI or client is sent to a hosted copilot endpoint on Azure Container Apps. The query is then routed to various processing services in an orchestrated RAG workflow that generates the relevant response - which is returned to the user through the frontend client.

The orchestration involves three processing services. The Azure AI Search service manages product indexes for the retailer's catalog, returning matching items for user queries based on similarity search and semantic ranking. The Azure CosmosDB service maintains a noSQL database with customer order history, returning product recommendations based on prior purchases. The Azure OpenAI service manages model deployments for embeddings, chat and evaluation - key capabilities required to implement the query vectortization, response generation, and quality assessment steps of our end-to-end application lifecycle.

Learning Outcomes

Implement a RAG-based copilot end-to-end with Prompty and Azure AI Studio

• Understand the RAG design pattern and Azure AI app architecture
• Provision and manage Azure infrastructure using azd-enabled AI templates
• Ideate and iterate on the application prototype using Prompty assets and tooling
• Evaluate and trace application execution for observability in cost and performance
• Deploy the custom copilot to Azure Container Apps for real-world usage
• Customize the sample with your data and extend it to meet your app scenariosx

Technology Used

[!IMPORTANT] The table lists all the tools and technologies used in the Contoso Chat sample. It is under active development and features some tools that are in preview - so anticipate regular updates.

Additional Resources and Continued Learning

[!TIP] If you are a workshop instructor or proctors, start with the session delivery resources.

Content Owners

Responsible AI

Microsoft is committed to helping our customers use our AI products responsibly, sharing our learnings, and building trust-based partnerships through tools like Transparency Notes and Impact Assessments. Many of these resources can be found at https://aka.ms/RAI. Microsoft’s approach to responsible AI is grounded in our AI principles of fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability.

Large-scale natural language, image, and speech models - like the ones used in this sample - can potentially behave in ways that are unfair, unreliable, or offensive, in turn causing harms. Please consult the Azure OpenAI service Transparency note to be informed about risks and limitations.

The recommended approach to mitigating these risks is to include a safety system in your architecture that can detect and prevent harmful behavior. Azure AI Content Safety provides an independent layer of protection, able to detect harmful user-generated and AI-generated content in applications and services. Azure AI Content Safety includes text and image APIs that allow you to detect material that is harmful. We also have an interactive Content Safety Studio that allows you to view, explore and try out sample code for detecting harmful content across different modalities. The following quickstart documentation guides you through making requests to the service.

Another aspect to take into account is the overall application performance. With multi-modal and multi-models applications, we consider performance to mean that the system performs as you and your users expect, including not generating harmful outputs. It's important to assess the performance of your overall application using generation quality and risk and safety metrics.

You can evaluate your AI application in your development environment using the prompt flow SDK. Given either a test dataset or a target, your generative AI application generations are quantitatively measured with built-in evaluators or custom evaluators of your choice. To get started with the prompt flow sdk to evaluate your system, you can follow the quickstart guide. Once you execute an evaluation run, you can visualize the results in Azure AI Studio.