Browse All Posts (457)
Dan Hellem announces general availability of Copilot Code Review for Azure Repos and explains how to enable and configure it. Teams can request or automate pull request reviews, apply suggested changes, and track usage-based costs through Azure, while retaining responsibility for approving code and merging changes.
David Ortinau compares a .NET MAUI app with platform-specific C# interfaces rebuilt using GitHub Copilot. His measurements show that AI reduced the effort of building a second UI, but maintenance and verification still cost more. Startup and package-size results also depended on runtime settings and application dependencies.
Anastasiya Tarnouskaya, Tucker Burns, and Vicente Rivera introduce Windows ML 2.9.42 Experimental and its MLIR-based DirectX Compute Graph Compiler backend. They explain model compilation, GPU acceleration and fallbacks, expanded Task APIs, and local-model serving, including GitHub Copilot CLI usage, while outlining current hardware limitations and the supported production path.
In this session, dotnet demonstrates Async Rx.NET and discusses the work underway to make it ready for use. Support for async/await first appeared in 2023 but has remained in preview with few changes; the session looks at the project’s renewed progress.
In this session from dotnet, Orleans provides a way to separate application structure from deployment topology. Drawing on experience with game backends and other services, the session explores scaling from a single process to multiple clusters and regions, including the effects of latency, communication costs, and operational complexity.
dotnet explores the decisions behind AI-powered developer workflows, from local versus cloud models to IDE, terminal, and browser environments. The discussion covers grounding agents with context, delegating work to sub-agents, balancing privacy and cost, and validating generated work through guardrails and human review.
dotnet shares production lessons from converting a large ASP.NET Web Forms UI with more than 180 components to Blazor WebAssembly. The talk covers migration strategy, authentication, architectural design, and performance pitfalls, discussing where Blazor met expectations and where practical solutions were needed.
dotnet explores building AI applications with .NET 11, EF Core 11, and Oracle AI Database 26ai. Demonstrations cover vector search and indexes, connecting Oracle vector storage to Microsoft Agent Framework, and using Oracle Deep Data Security to provide least-privilege access for AI agents without changing application code.
dotnet explores ASP.NET Core 11's built-in OpenTelemetry metrics and traces and how they follow HTTP server semantic conventions. The session explains these conventions and shows how to validate telemetry and create consistent dashboards and alerts across services written in different programming languages.
dotnet introduces infrastructure as code in plain C# using Pulumi's open-source .NET SDK. The session covers building and deploying Azure infrastructure, then using C# 15 and .NET 11 to catch infrastructure mistakes at compile time, including when AI agents generate the changes.
In this dotnet talk, three experimental projects explore bringing modern web tools to .NET: React and JSX bindings for ASP.NET MVC, C# support for Astro, and a TypeScript compiler targeting the CLR. The session covers their designs and how to bridge platforms that do not ordinarily work together.
dotnet demonstrates how to turn a basic .NET console application into a maintainable terminal UI. The live build covers prompts, validation, tables, panels, and progress feedback, alongside separating presentation from application logic, testing workflows without terminal automation, and using Aspire for observability and diagnostics.
dotnet presents a session on building AI-enabled Blazor applications with Telerik tooling and Microsoft Agent Framework. It covers development assistance through MCP and AI skills, then explores UI components and protocols that let agents collaborate with users through dynamic, adaptive interfaces.
In this session, dotnet explores how Bicep and Terraform modules can support resilient, reusable shared cloud infrastructure. The session covers module authoring, versioning, and distribution, along with using modules to enforce naming conventions, tagging standards, and organizational policies as infrastructure needs grow.
In this talk from dotnet, the presenter explores miniservices as a middle ground between simpler applications and full-blown microservices. The approach uses Azure databases, App Service, Azure Functions, and Service Bus to reduce architectural complexity for small and medium-sized businesses without requiring a Kubernetes cluster.
dotnet explores building adaptive dual-screen experiences with Avalonia and TwoPaneView. Practical samples cover list-detail interfaces, navigation and editors, showing how layouts respond to folding, spanning and single-screen configurations while preserving application state, keyboard focus and the user’s context across mobile and desktop devices.
dotnet explores SE.Redis, the library used to access Redis from .NET applications and caching tools. The video covers client-side caching and active-active capabilities, then examines how SE.Redis 4.0 uses newer C# and .NET features to provide a more efficient API while retaining support for the existing API.
In this dotnet video, a T-SQL Analyzer extension connects a thin TypeScript front-end to a persistent C# backend. The talk covers stdin/stdout JSON communication, process supervision, NuGet distribution through dnx, and live editor diagnostics, showing how an existing .NET library can power a VS Code extension.
Fokko at Work demonstrates selected GitHub Copilot features in Visual Studio Code 1.147, including model routers, native visuals in agent responses, saved drafts, and browser use. The video also covers request scheduling and model comparison improvements, with a reminder that availability can depend on enterprise policies and pricing plans.
Manoj Bableshwar announces general availability of Managed Compute in Microsoft Foundry for serving open models on dedicated GPUs. He walks through Python deployment, scaling, prompt caching, and supply-chain protections, and introduces preview support for LoRA adapters and custom model weights.
Maria Victoria Gromysch shares how NVIDIA combines GitHub Copilot, GitHub Actions, and domain-specific AI agents to address hardware reliability. The session covers test generation, code review, CI/CD validation, and training agents on internal diagnostics to handle hardware constraints and firmware interactions that general-purpose models can miss.
Shree_Divya_M_V introduces the MSMQ built-in connector for Azure Logic Apps Standard, which connects workflows to existing Windows messaging environments through a privately networked broker. The walkthrough covers installation, queue configuration, sending and receiving messages, and public-preview limitations, including duplicate delivery, message-size constraints, and unsupported ordered delivery.
Allison explains how to choose the repositories shown in GitHub’s new dashboard experience. The Favorite repositories section syncs with pinned repositories in GitHub Mobile after a refresh, with a shared limit of 25 repositories. Users can search, select, and save favorites through the repository picker.
Shree_Divya_M_V announces general availability of SQL as a runtime storage provider for Hybrid Logic Apps, excluding regular Azure-hosted Logic Apps Standard. The article explains runtime-owned schemas, customer responsibilities, and SQL Server operational practices for keeping workflow state on customer-managed infrastructure.
Shree_Divya_M_V introduces semi-connected Integration Account support for Azure Logic Apps Hybrid. The capability caches supported artifacts in local SQL storage so workflows can continue during temporary Azure connectivity outages. The article covers configuration, validation, synchronization requirements, and limitations for operations that need live access to certificate keys.
Cybersecurity researcher Niels Provos examines how autonomous AI agents can be tricked into leaking private data while browsing the web. This GitHub interview clip describes a honeypot website with hidden prompt injections and highlights why agent workflows need strict security boundaries.
goupadhy outlines the November 2026 on-premises data gateway release, version 3000.342. The update adds Dual Relay resiliency, addresses a cryptography dependency vulnerability, and improves authentication, memory use, and request processing. It also aligns gateway query execution with the November 2026 release of Power BI Desktop.
Anders Zwartjes and Kushmeen Kambow demonstrate how Foundry Local on Azure Local supports AI workloads in customer-controlled environments. The session covers model deployment, evaluation, agents, retrieval, and scalable inference, with patterns for disconnected operations, multi-GPU inference, and evaluating model quality without data leaving the environment.
Wade Callison and Max Pawlikowski share lessons from TowneBank’s use of AI agents for banking file transfers. The session covers how Azure API Management, Azure Logic Apps, and Files.com support natural-language troubleshooting, customer provisioning, and guided workflow setup, including the context agents need to act.
Krishna Sunkavalli and James Brotsos explain how Microsoft Defender helps secure AI runtime using a Zero Trust approach. This Microsoft Ignite session covers reducing agent risk and defending against prompt injection, tool misuse, and identity abuse, with security guidance mapped to OWASP and MITRE.
Michael Draper, Brandon Bray, and Mariam Gewida cover how applications can prepare for evolving Windows security standards. The session addresses baseline protection, NTLM-to-Kerberos migration, Admin Protection, code signing, and service deprivileging, with a look at how Adobe is adopting some of these technologies.
Troy Lainhoff and Vladimir Suarez share what Microsoft experts are seeing on the frontlines of cyber defense. This Microsoft Ignite session explores practical guidance for making threat intelligence actionable, operationalizing AI agents, and building cyber resilience in response to current threats and challenges.
Rafsan Huseynov and Matthew Devaney demonstrate how to bring unmanaged AI agents under Microsoft identity and governance controls. The session covers Entra Agent IDs, Conditional Access, lifecycle workflows, access packages, and Purview policies, including how an external Python agent joins the same control plane through the Agent 365 SDK.
Greg Schellenberg, Alex Klausner, Nuno Costa, and Kiran Kumar Dowluru demonstrate how Microsoft Security Exposure Management maps threat intelligence to exposures across a digital estate. This advanced session covers validating findings and producing an agent-assisted, risk-ranked action plan to help security teams prioritize remediation.
Tanmay Ganacharya and Yoni Nave explore how Microsoft combines threat visibility, expert-led research, and AI to turn intelligence into defensive action. This Microsoft Ignite session includes demos and practical strategies for addressing current threats, moving beyond static threat intelligence feeds toward actionable protection.
Harald Kirschner and Ross Wollman explore how teams can tailor GitHub Copilot to their development workflows through team instructions, tool connections, and delegated tasks. The session covers how testing, code review, and organizational policies help keep developers in control as AI agents take on more work.
Leslie Welch and Kevin Pereira explain how data engineers and BI developers can use GitHub Copilot in Microsoft Fabric workflows. This practical session introduces AI-assisted coding and testing to experienced developers, focusing on where AI adds value and how to reduce repetitive work when building data solutions.
Anastasiya Tarnouskaya, Pawan Tahiliani, and Daniel Freedman cover deploying custom and open-source AI models across enterprise PCs with Windows ML. The session examines moving from pilot to production and using one framework for high-performance local inference across CPU, GPU, and NPU hardware.
Danielle Augustin, Sharon Sandhu, and Etan Basseri explain how a unified policy engine combines identity, network, device, and risk signals to make access decisions. This advanced Microsoft Ignite session explores managing policy dependencies and scaling adaptive access for humans and AI agents with confidence and control.
Nicholas Balestrino, Zohar Raz, Ryan Jansen, and Austin Laugesen cover deploying and operating Power Platform and Copilot Managed Runtime solutions at enterprise scale. The session explores environment strategies, application lifecycle management, security, monitoring, and administration for apps, AI agents, and workflows built with Power Apps, Power Automate, and Copilot Studio.