Udemy
    •  
    •  
    •  
    •  
    •  
    •  
    •  
    •  
Turn what you know into an opportunity and reach millions around the world.
Learn More
Your cart is empty.
Keep shopping
Build a Customer Support Agent using OpenAI and AzureML
Rating: 4.2 out of 5(6 ratings)
928 students

Build a Customer Support Agent using OpenAI and AzureML

Master the end-to-end process of building, training, and deploying an AI-powered customer support assistant using Azure
Created byProjectPro Academy
Last updated 10/2025
English
English [Auto],

What you'll learn

  • Recognize the challenges in managing high volumes of customer support tickets and the need for automation
  • Learn to set up and configure an Azure ML workspace for managing and tracking machine learning projects
  • Implement and run AI workflows using Azure ML pipelines
  • Explore how to use Open AI Embeddings to convert text data into embeddings for efficient retrieval
  • Understand the role of vector databases in storing and retrieving embeddings
  • Implement a vector database to store and retrieve text embeddings efficiently
  • Learn the Retrieval-Augmented Generation (RAG) principles and how it combines retrieval and generation techniques
  • Understand how to generate responses using LLMs based on retrieved information
  • Explore techniques for generating multiple response candidates and selecting the best one
  • Build a user-friendly demo application using Streamlit to showcase the AI assistant's capabilities
  • Create a Feedback loop to improve the response based on prompt improvements
  • Learn the steps involved in deploying the AI application on Azure

Course content

6 sections • 10 lectures • 1h 59m total length

  • Project Overview5:35

    Explore ai-assisted customer support with generative ai and lms on OpenAI and Azure, analyzing and categorizing tickets by urgency and sentiment, routing to teams, and generating responses with predefined templates.

  • Approach8:02

    Define an end-to-end customer support app using Python, Streamlit, and GPT-4 with rag to generate dynamic responses from a knowledge base. Prioritize data quality, privacy, and human-in-the-loop safeguards.

Requirements

  • Basic Python programming knowledge and API usage understanding.
  • Familiarity with machine learning fundamentals and cloud concepts.
  • Access to an Azure account and a working local development environment.

Description

This project aims to enhance customer support efficiency and reduce operational costs by leveraging Large Language Models (LLMs) and Azure Machine Learning for automated ticket categorization, prioritization, and response generation.

1. Introduction to AI-Powered Customer Support Automation

Begin with an overview of the challenges in managing large volumes of customer support tickets and the growing importance of automation. Understand how AI technologies like Azure ML and OpenAI can transform traditional customer support systems into intelligent, responsive agents.

2. Azure ML Workspace Setup and Data Analysis

Learn how to set up and configure your Azure ML workspace, connecting it seamlessly with your local development environment. You’ll then load and analyze a retail dataset containing customer support tickets to identify patterns and insights that will guide your model development.

3. LLM Integration and Vector Database Implementation

Integrate a pre-trained Large Language Model (LLM) to generate embeddings and responses. You’ll then set up a vector database using FAISS to store these embeddings efficiently, enabling fast and relevant retrieval of context-based information for customer queries.

4. Prompt Engineering and RAG Architecture

Master prompt engineering to design and refine input prompts that yield precise and contextually relevant responses. Implement the Retrieval-Augmented Generation (RAG) framework, combining retrieval-based and generative techniques to ensure your AI assistant responds intelligently using the stored vector data.

5. Response Generation, Sampling, and Feedback Loop

Develop robust response generation logic using the LLM and retrieved data. Implement response sampling to produce multiple candidate answers and select the most suitable one. Establish a feedback loop to continuously improve prompts and responses based on user interactions.

6. Streamlit UI Development and Azure Deployment

Focus on code modularity to maintain clarity and scalability. Build an interactive Streamlit interface to showcase your AI support agent’s capabilities. Finally, deploy your application on Azure ML, ensuring it operates efficiently, scales with demand, and remains easy to maintain in production environments.

Who this course is for:

  • Data scientists and AI engineers eager to apply LLMs in real-world scenarios.
  • Developers seeking to automate and enhance customer support operations.
  • Cloud and ML professionals aiming to integrate Azure ML with OpenAI for production-grade AI solutions.

Report abuse