Building an Interactive Chatbot for Microsoft NAB 2025 Sports Business Hub with Azure AI

Check out the live site at: http://aka.ms/4Sports

Introduction

For the NAB Show 2025, SOUTHWORKS was commissioned to develop a solution utilizing Azure AI that would enable NAB attendees visiting the Microsoft "Reimagining Sports Media Through AI-Powered Transformation" activation in the Sports Business Hub to navigate and discover content through a natural language chatbot experience. The objective was to illustrate how Microsoft partners utilize AI to enhance various facets of their businesses, ranging from operations to fan engagement. This blog post delves into our solution and demonstrates how the capabilities of Azure AI Foundry made it incredibly easy to build a solution quickly and efficiently.

The Challenge

SOUTHWORKS successfully tackled the challenge of developing an AI chatbot within a streamlined timeframe, grounded with knowledge about Microsoft and the solutions its customers and partners were leveraging for the sports industry, enabling it to swiftly answer related questions. The knowledge source was designed to be expandable and modifiable, drawing from various data sources such as web pages, PDFs, and video assets.

The Solution: Microsoft AI-Powered Chatbot

To address these challenges, we employed various Microsoft AI resources:

• Azure AI Foundry: Provided a scalable framework for developing and deploying AI models, streamlining and accelerating the integration of AI into custom workflows.
• Azure OpenAI Service: Leveraged GTP models for generating context-aware responses to user queries, enhancing conversational AI capabilities.
• Azure AI Search:  Utilized for high-performance indexing and retrieval, enabling efficient access to relevant data with advanced search capabilities.
• Azure AI Content Understanding: Used to analyze video content transforming it into structured, organized, and searchable data, making it possible to query video assets without any model training or fine-tuning.

‍

The diagram below describes the high-level architecture of the solution deployed in Azure and the Azure AI services used for both the user query path and the indexing process.

‍

User Flow: Enhancing Attendee Experience with Azure AI

The user query flow is implemented with a traditional RAG pattern. When a user submits a message through the chat interface (hosted on a static site), the API driven backend handles the request. It first fetches recent chat history from Azure Cache for Redis (acting as a memory store) to maintain conversation context. The query is then sent to Azure OpenAI Service, where GPT-4o processes the input and generates a relevant, natural-sounding response.

To enhance answers with more grounded knowledge, the system uses Azure AI Search as a vector store. This allows the system to perform semantic search over indexed content, bring the results into the context of the LLM, improving responses accuracy and relevance.

Content Ingestion Flow: Streamlining Content Management with Azure AI

The Microsoft AI Assistant for the NAB Show 2025 Sports Business Hub chatbot uses a streamlined ingestion pipeline to manage and index diverse content types like articles and videos, leveraging Azure AI services throughout the process.

It starts when a back-office user runs the Content Ingestion app, which reads a JSON file containing URLs to process. These URLs point to the content that needs to be ingested.

To extract meaningful data from each piece of content, the app combines a scraping library to pull structured content from web pages, with Azure AI Content Understanding to analyze video content.

Next, the extracted content and metadata is processed with GPT-4o, which generates concise synopses based on a predefined prompt. For each piece of content, a JSON file is created and uploaded to Azure Blob Storage.

These files are then indexed using the “Import and vectorize data” feature in Azure AI Search. The JSON documents are vectorized using the Azure OpenAI Service embeddings model, making them searchable via semantic search.

Finally, the new Azure AI Search index is connected as a data source to Azure OpenAI Service, enabling GPT-4o to use this knowledge base when answering user questions—ensuring responses are both relevant and grounded in the latest indexed material.

Conclusion

By harnessing Azure AI Foundry and Azure AI services, SOUTHWORKS seamlessly implemented a chatbot experience grounded with relevant data about Microsoft and their sports activation at the NAB Show 2025 Sports Business Hub.

Enhancing Attendee Experience & Content Management with Azure AI

The Microsoft sports activation at the NAB Show 2025 Sports Business Hub showcases how Azure AI can transform attendee experience, fan engagement and content management. With Azure OpenAI Service, Azure AI Search, and Azure AI Content Understanding, the chatbot delivers context-aware interactions while streamlining content ingestion and retrieval. These technologies ensure that attendees stay informed, engaged, and connected to their favorite sports like never before.

‍