Browse Artificial Intelligence Videos (194)

John Maeda explains why agentic systems benefit from a “clean molt”: trimming unnecessary context so the system doesn’t accumulate noise that degrades results. The tip focuses on practical context management—what to keep, what to drop, and why “more context” can make outputs worse.

GitHub shows how GitHub Copilot CLI can scan a repository and generate a pull request that follows contribution guidelines, issue templates, and team rules—reducing the manual work of formatting and filling out PR details.

Visual Studio Code briefly introduces Agent Plugins in VS Code, explaining that you can install plugins to add agent skills, commands, MCP servers, hooks, and more.

Rob Bos covers GitHub Copilot’s token-based billing, focusing on what “tokens” mean in practice and how usage-based pricing can affect Copilot costs for teams and organizations.

Microsoft Developer hosts a Cosmos DB Conf 2026 session where Sergiy Smyrnov demonstrates migrating an AdventureWorks-based ASP.NET/EF Core app from a relational database to Azure Cosmos DB for NoSQL, using GitHub Copilot and Cosmos DB Agent Kit prompts to plan the move and rewrite the data layer.

Microsoft Developer shares Yohan Lasorsa’s session on why “vector search by default” often fails in RAG, and how to improve retrieval quality by using agentic RAG with Azure Cosmos DB hybrid search (vector + full-text) so an AI agent fetches the right context when it actually needs it.

Microsoft Developer shares Mick Feller’s customer story on how Office Depot (ODP Corporation) built a homegrown enterprise AI “Personal Assistant” backed by Azure Cosmos DB, focusing on time-series + HR profile context, autoscale-driven growth, and cost/performance control.

Microsoft Developer features Sajeetharan Sinnathurai explaining how to set up a practical Azure Cosmos DB development environment (VS Code, emulator, testing, and CI/CD considerations) and how to use AI coding assistants more effectively by providing schema context and better prompts.

Microsoft Developer shares a keynote interview with Vercel CEO Guillermo Rauch on how AI agents change app design, why serverless compute plus serverless data matters, and what Azure Cosmos DB’s RU-based query economics look like as tokens and agent-driven workloads grow.

Authorised Territory demonstrates how to define a skill in code and attach it to an AI agent built with the .NET Agent Framework, using a locally running LLM via Ollama.

Visual Studio Code features Reynald Adolphe demonstrating how to create and use Custom Agents in VS Code (Agent Mode) so GitHub Copilot can take on specialized roles—like a security reviewer—using your project context, tools, and workflow to produce more focused results.

Visual Studio Code features Reynald Adolphe explaining how Agent Skills in VS Code package instructions, scripts, and resources into reusable workflows, with a quick demo of creating and modifying a skill to automate multi-step tasks like updating docs and generating prompts.

Visual Studio Code features Reynald Adolphe using GitHub Copilot in VS Code to explain and compare customization features—Custom Instructions, Prompt Files, Agent Skills, Custom Agents, and Hooks—plus ways to learn them via charts, quizzes, and scenario-based references.

Visual Studio Code features Reynald Adolphe explaining GitHub Copilot Hooks in VS Code: how to run commands automatically at specific lifecycle events during an agent session to keep formatting, validations, and scripts consistent without manual steps.

Visual Studio Code shows how to use Prompt Files in VS Code to stop rewriting the same GitHub Copilot instructions, making repeatable workflows (like quizzing open files or simplifying code) easier to run and share across projects and teams.

Visual Studio Code features Reynald Adolphe explaining how GitHub Copilot Custom Instructions in VS Code work in practice, showing how to create reusable “rulebooks” that steer Copilot toward your coding standards, conventions, and preferences (including SOLID and accessibility) without repeating guidance every time.

Visual Studio Code shows how to customize AI in VS Code using agent-based building blocks—agents, skills, instructions, prompt files, and hooks—so teams can reuse prompts, enforce standards, and streamline common development tasks.

Visual Studio Code shows Reynald building a “Repo Analyzer” app in VS Code using GitHub Copilot customization features—Prompt Files, Custom Instructions, Agent Skills, Custom Agents, and Hooks—to enforce repo standards, update docs, and streamline coding tasks in one workflow.

Microsoft Developer shares John Maeda’s SXSW tip for getting better results from AI agents: keep a smaller, working replica of your app that fits within a model’s context window so you can iterate on prompts and changes more effectively.

Microsoft Developer shares John Maeda’s SXSW tip on using a “tuned harness” to keep AI models on track, from prompting through tool calling, and points to tools like GitHub Copilot CLI (and the Copilot SDK) to help turn interesting outputs into something reliably useful.