MsftTechNews

The best session catalog is an agent that understands what you are building. Build CLI flips it: it reads project dependencies, finds relevant Microsoft Build sessions, then helps you scaffold from what you learned. Less browsing. More building. github.com/microsoft/Buil…

Azure DevOps teams can now bring DevOps actions into VS Code Agent using MCP. Work items. PRs. Pipelines. Repos. Test plans. Wiki. Less tool switching. More flow. What Azure DevOps task would you want your agent to handle first? Read more: techcommunity.microsoft.com/blog/azuredevc…

@JayCee83519 @MSCloud @MSFTReactor Apologies for the confusion. Seems like the issue has been resolved.

@MsftTechNews @MSCloud @MSFTReactor

Azure Infra Summit 2026 is live today. Join for the stuff infra teams actually need: landing zones, IaC, AKS, networking, resiliency, storage, backup, cost, and production gotchas. Free. Virtual. Engineering-led. Register and watch live: aka.ms/MAIS-reg #Azure

AI agents won’t replace engineers. Coordination drag will. The hard part isn’t detection. It’s chasing vendors, validating fixes, & closing loops. Agents win when they carry context until work is done. What workflow is your biggest coordination debt? techcommunity.microsoft.com/blog/appsonazu…

The commit is no longer the atomic unit of delivery. In AgenticOps, the spec is. Agents write code. Evals gate it. Refusal rules bound it. Humans review intent. That is the shift Copilot teams need to understand. Read AKS-Lab-GitHubCopilot: techcommunity.microsoft.com/blog/azuredevc…

The UI is becoming the new API. Computer-using agents can automate work trapped inside portals, desktop apps, and legacy systems. Not by chatting. By doing the work. What legacy workflow would you automate first? #AIAgents Read More: techcommunity.microsoft.com/blog/copilot-s…

AI agent governance fails when review happens after damage. Agent 365 + Purview move controls to runtime: block labeled files, stop sensitive sends, log the trail. Can your team prove what each agent did? Learn more: techcommunity.microsoft.com/blog/microsoft…

AI agents don’t fail in demos. They fail in workflows. Microsoft’s Agent Evaluations preview points to the real bar: tests for coherence, grounding, and multi-turn tasks. Stop vibe-checking agents. Start scoring them. Try the CLI: devblogs.microsoft.com/microsoft365de…

Most cloud events talk to the buyer. Azure Infra Summit talks to the person who gets paged. 3 days on Build, Operate, Optimize for real Azure production work. Free, virtual, May 19–21. Register: developer.microsoft.com/en-us/reactor/… #Azure

Your agentic CI bill is probably hiding in “reads.” If an agent uses LLM turns to fetch diffs, labels, or comments, you’re paying for reasoning where none is needed. Move deterministic reads before the agent starts. Read the breakdown: github.blog/ai-and-ml/gith…

Authentication is not authorization for AI agents. Scoped, short-lived tokens tell your APIs what the agent may actually access. That boundary matters when agents are wrong, creative, or attacked. Try the Curity + Microsoft azd template. devblogs.microsoft.com/azure-sdk/azd-…

Agent sprawl isn’t a model problem. It’s an architecture problem. Without a control plane, every agent that calls tools, touches data, or burns tokens becomes shadow IT with autonomy. Read more: techcommunity.microsoft.com/blog/azurearch…

OpenClaw is growing because agent builders want control: tools, state, and long-running workflows—not prettier demos. Join OpenClaw: After Hours at GitHub HQ or Twitch on June 3. Register: github.blog/open-source/re…

AI security can’t live in a browser policy. Employees use AI, SaaS, and private apps from everywhere. The control point is identity: who’s asking, from what device, with what risk. Read the Entra update, then rethink your VPN. techcommunity.microsoft.com/blog/microsoft… #ZeroTrust

Image generation doesn’t need another tab. Foundry Toolkit v1.0.1 puts GPT-Image-2 discovery, deployment, prompting, and code export inside VS Code. Less context-switching. Faster experiments. Learn more: techcommunity.microsoft.com/blog/azuredevc…

MCP safety should not live in scattered if statements. If agents can read files, call APIs, & query databases, every tool call needs a policy gate. CTA: audit your MCP tool path today. 👉 devblogs.microsoft.com/dotnet/governi…

Bring your own model to Foundry Agent Service means more than model choice. Route through your existing gateway. Keep governance intact. Use compatible models across providers. Move faster without rebuilding controls. That’s enterprise AI architecture. techcommunity.microsoft.com/blog/azure-ai-…

GitHub’s Copilot billing change is a good sign: AI tools are growing up. @mariorod1 Agentic workflows need visibility, controls, and smarter scaling. The teams that win won’t use AI everywhere. They’ll know which workflows are worth automating. Read: github.blog/news-insights/…

Microsoft’s MAI models aren’t “new endpoints.” They’re Microsoft pulling speech, voice, and image AI deeper into Foundry. Less vendor patchwork. More native governance. Azure devs should rethink agent architecture now. Read: techcommunity.microsoft.com/blog/appsonazu…