Use VS Code Agents, GitHub Copilot, and the MSSQL extension to build AI apps | Data Exposed

Carlos Robles demonstrates how to use VS Code Agents with GitHub Copilot and the MSSQL extension to build an AI app that uses semantic (vector) search, including an agent-driven workflow for proposing and applying changes and working with SQL schema design inside VS Code.

Overview

The video walks through an end-to-end developer workflow for building an AI-enabled application using:

• VS Code Agents (including a “plan mode” workflow)
• GitHub Copilot inside VS Code
• MSSQL extension for VS Code for working with SQL databases
• Semantic search / vector search as the app capability being implemented

What the episode covers (chapter-based)

Scenario overview and plan mode

• Introduces the scenario and how an agent-driven workflow can be used to plan changes before applying them.

New Agents experience in VS Code

• Shows the updated Agents experience in VS Code and how it fits into the development loop.

Using an OpenAPI spec in the workflow

• Demonstrates how the running app used an OpenAPI spec.
• Walks through an agent flow to:
  • Propose changes to the spec
  • Apply the proposed changes

Investigating and updating an issue related to vector search

• Uses the agent workflow to investigate an issue and update it in the context of implementing semantic/vector search.

MSSQL extension and Dev Containers

• Highlights using the MSSQL extension in VS Code.
• Covers using Dev Containers as part of the development environment.

Schema designer with GitHub Copilot integration

• Demonstrates the schema designer experience and how GitHub Copilot integrates into that workflow.

Resources

• MSSQL extension for VS Code: https://github.com/microsoft/vscode-mssql

People and links mentioned

• Anna Hoffman (Twitter): https://twitter.com/AnalyticAnna
• AzureSQL (Twitter): https://aka.ms/azuresqltw
• Data Exposed playlist: https://aka.ms/dataexposedyt