Ship an AI Agent without leaving VS Code

Julia Muiruri demonstrates an end-to-end workflow for building and shipping an AI agent directly from Visual Studio Code, using GitHub Copilot and the Foundry Toolkit extension to pick a model, test the agent against a real product catalog, deploy to Microsoft Foundry, and connect it to a website via a secure proxy.

Overview

The video walks through a practical demo of creating an AI agent inside VS Code and taking it all the way to deployment.

Tools and platform used

• Visual Studio Code
• GitHub Copilot
• Foundry Toolkit VS Code extension
• Azure AI Foundry (referred to as “Microsoft Foundry” in the description)

Demo scenario: game recommendations from a real catalog

• The agent is built to recommend games.
• Recommendations are grounded in a real product catalog.

Build, test, and deploy flow

• Choose a model using the Foundry Toolkit extension.
• Build and test the agent within VS Code.
• Deploy the agent to Azure AI Foundry.
• Connect the deployed agent to a website.
• Use a secure proxy for the website-to-agent connection.

Video chapters

• 00:00 Introduction
• 00:38 Demo
• 09:30 Recap