Chatbot Playground App with AKS KAITO and MS Foundry Hosted LLMs

What does it look like to test multiple AI providers and deployed LLMs from one app?

I walk through my Streamlit chatbot playground built with Microsoft Agent Framework and deployed on Azure Kubernetes Service. The purpose of this app is a testing playground/harness to compare model behavior across Azure AI Foundry and AKS-hosted KAITO, and show how this app helps test prompts, providers, and deployed models from a single interface.

I also explain how the chatbot itself is deployed into the same AKS cluster, making it a practical setup for teams that want a reusable internal AI testing tool. I will continue to add features and capabilities to monitor, diagnose and demo AI Agent various capabilities such as MCP integration, tool calling, triggering skills, context & memory management and more.

Find this app's code in my github repo https://github.com/RoyKimYYZ/azure-ai-apps/tree/main/agentframework
This GitHub repo is meant to be extended. You can use it as a starting point to add your own AI providers, models, prompts, and agents.

Intro 0:00
Streamlit Chatbot - Microsoft Foundry 0:17
Streamlit Chatbot - AKS KAITO 3:29
Architecture 6:50
Application Code & Configuration 7:38

Blog: www.roykim.ca
X: @RoyKimYYZ

Видео Chatbot Playground App with AKS KAITO and MS Foundry Hosted LLMs канала Roy Kim