Microsoft Mechanics Podcast

Match AI-powered attacks at machine speed with Project Perception. Point red, blue, and green AI agent teams at a known threat actor, confirm which vulnerabilities attackers can reach, then deploy KQL detections, harden your posture, and merge a tested code fix in minutes with Microsoft Defender. Stay in control and approve key actions before agents run them. Run Microsoft's pre-built playbooks or build your own, and manage sessions from the Defender portal or from your terminal with the MCP plug-in. Keep token costs low with MAI Cyber Flash, a purpose-built security model inside MDASH. Dave Weston, CVP Microsoft Security, shares how Microsoft's threat intelligence and incident response expertise power agents you direct.  👥 Who it's for: SOC analysts, security engineers, and red and blue teamers evaluating agentic AI for defense, plus AppSec and DevSecOps teams who own vulnerability remediation.  ⏱️ Chapters: 00:00 Fight AI-powered attacks with AI agents 00:45 How AI attacks adapt mid-breach 01:39 Jailbroken and abliterated models 02:04 Inside the Project Perception security stack 03:28 Red, blue, and green agent teams 04:07 Playbooks that orchestrate agent workflows 04:43 Ask Perception if you're protected 06:06 Red team agents emulate a threat actor 06:59 Approve high-impact agent actions 07:26 MDASH findings confirm reachable vulnerabilities 08:22 Blue team agent writes KQL detections 08:54 WAF, auth check, and tested code fix 09:36 Pull request merged into production 10:14 Session reports and MCP terminal plugin 11:02 MAI Cyber Flash keeps token costs low 11:57 Join the Project Perception preview AI helps attackers find, weaponize, and exploit vulnerabilities at a speed and scale no human team can match. Jailbroken and abliterated models generate custom tooling that adapts mid-breach, leaving defenses that hunt for known techniques chasing a target that changes every time. Project Perception answers with a full agentic security stack. It starts with trillions of daily security signals, correlates them in the Microsoft security graph, and uses multiple AI models, including purpose-built ones, to reason over risk and prioritize what matters most. Specialized agents act on that reasoning: red team agents simulate attacks and prove where you're vulnerable, blue team agents investigate and build detections, and green team agents harden posture and remediate. Playbooks orchestrate them through proven SOC workflows, and codified actuators let agents perform the fix instead of only recommending it. In the demo, one prompt, "Is my organization protected against Onyx Sleet?", launches threat intelligence, reconnaissance, and pen test agents. They pull MDASH code-scanning findings and confirm an internet-exposed SQL injection. A detection authoring agent writes KQL detections to cover the gap, and the posture remediation agent shields the app with a web application firewall, adds an authorization check, and produces a tested code fix merged through a pull request. Agents stop and ask before any high-impact action, so you stay in control. Every session produces a shareable report with the attack path, remediation plan, and timeline. An MCP tool plugin runs the same work from your terminal, and MAI Cyber Flash, a dedicated cybersecurity model inside MDASH, handles most vulnerability work to keep token costs low.  🔗 Related links:  ► Get started at https://microsoft.com/perception ► Unfamiliar with Microsoft Mechanics? Microsoft's Official Video Series for IT — Subscribe https://www.youtube.com/c/MicrosoftMechanicsSeries — Microsoft Tech Community: https://techcommunity.microsoft.com/t5/microsoft-mechanics-blog/bg-p/MicrosoftMechanicsBlog — Podcast: https://microsoftmechanics.libsyn.com/podcast ►Join us on social: — https://twitter.com/MSFTMechanics —  https://www.linkedin.com/company/microsoft-mechanics/ — https://www.instagram.com/msftmechanics/ — https://www.tiktok.com/@msftmechanics 

Take a business process from prompt to production by building agents, workflows, and apps in one place with Copilot Studio. Describe the agent you need and generate its instructions and reusable skills, choose the model, connect MCP tools and specialist agents, then test its quality at scale with Evaluations. Trigger workflows from incoming email, classify submissions, hand work to your agents, and pause for human review on the decisions that need it. Track timing and paths for every step in the Activity tab. Then generate a full stack app on your Dataverse data so your team can filter, review, and approve every bid from one shared view. Jack Rowbotham, Copilot Studio Senior Product Manager, shares how to automate a procurement process for bids from multiple vendors, using agents that reason, route, and recommend.  👥 Who it's for: IT admins, Power Platform makers, solution architects, and developers building agentic solutions, plus operations and procurement teams automating reviews and approvals. ⏱️ Chapters: 00:00 Build agents, workflows, and apps in one place 01:04 Automate a multi-vendor bid review end to end 02:18 Build an agent from a prompt 03:28 Add models, MCP tools, and specialist agents 04:13 Test the bid agent in Preview 05:05 Evaluations 05:59 Automated workflows 07:42 Track every run in the Activity tab 08:13 Build an app from a prompt 09:10 Approve bids from the team app 09:37 Get started with Copilot Studio  Copilot Studio brings agents, workflows, and apps together in one place, all running on the GitHub Copilot harness. In this video, a supplier email triggers a workflow, a bid evaluation agent checks the bid against tender requirements, and the procurement team makes the final call in a shared app. Describe the agent you want, including the requirements to assess, the logic to apply, and the live data source to connect to. Copilot Studio reasons over your prompt, maps out its build steps, and generates the agent's name, instructions, and skills for requirement coverage and evidence verification. Connect Work IQ for supplier emails and the Dataverse MCP server for bid data, keep Claude Sonnet as the reasoning model, ground the agent in an RFP example PDF, and add a Supplier Risk specialist agent. Test the agent in the Preview tab and watch its chain of thought as it flags gaps in certification, cold-chain handling, and insurance minimums. Then run Evaluations with your own conversations, auto-generated tests, or a CSV file. Here, 18 test cases met the evaluation criteria 89% of the time. In the Workflows tab, drag in an Outlook "When a new email arrives" trigger, a Classify step for RFP submissions, a Get Attachment connector, and an Agent step running the Bid Evaluation Agent. An if/else Human Review step brings in a person on flagged bids, and the Researcher agent and a recommendation agent finish the job before results land in Dataverse. Finish in Apps by describing a Bid Management App connected to your Dataverse table, or to sources like SharePoint through connectors. Apply company branding with a follow-up prompt, then filter bids that need human review and approve decisions from the app.  Try it for yourself at copilotstudio.microsoft.com. Subscribe to Microsoft Mechanics for more videos like this. 🔗 Related links: — Get started with Copilot Studio: https://copilotstudio.microsoft.com ► Unfamiliar with Microsoft Mechanics? Microsoft's Official Video Series for IT — Subscribe: https://www.youtube.com/c/MicrosoftMechanicsSeries — Microsoft Tech Community: https://techcommunity.microsoft.com/t5/microsoft-mechanics-blog/bg-p/MicrosoftMechanicsBlog — Podcast: https://microsoftmechanics.libsyn.com/podcast  ► Join us on social: — https://twitter.com/MSFTMechanics — https://www.linkedin.com/company/microsoft-mechanics/ — https://www.instagram.com/msftmechanics/ — https://www.tiktok.com/@msftmechanics

Build a specialist AI agent without writing a line of code. Describe what you want in one prompt, and Copilot Agent Builder generates the name, instructions, skills, and knowledge setup for you — grounded only in the SharePoint sites, files, and Copilot Connectors you choose. Teach your agent a Skill to handle recurring work, like drafting weekly status emails from your own template, then test it in Preview before you publish. Share it with people or groups and set exact permissions, track real usage in the new Monitor tab, and push your best agents into your org's Agent Store catalog. Ned Friend, Group Product Manager for Copilot Agent Builder shares how to turn an agent into a tool your whole team can run, measure, and trust. 👥 Who it's for: Team leads and managers who want trusted, repeatable answers for their team, plus IT admins and Microsoft 365 power users building specialist Copilot agents without code. ⏱️ Chapters: 0:00 Build specialist agents inside Copilot 0:34 Ground agents in trusted org knowledge 1:31 Create a new hire onboarding agent 2:15 Natural-language agent creation 2:41 Ground answers in SharePoint knowledge 4:13 Skills for repeatable tasks 4:53 Precise knowledge grounding 6:37 Preview and test before publishing 7:03 Share your agent with set permissions 7:39 See the new hire's agent view 8:22 Usage monitoring 8:41 Publish agents to your org catalog Copilot Agent Builder lets you create a specialist agent by describing what you want it to do in plain language. Agent Builder interprets that prompt and automatically generates the agent's name, description, instructions, skills, knowledge sources, and suggested prompts, which you can review and edit directly. Ground every answer in the exact knowledge you choose: SharePoint and OneDrive files, Outlook email and calendar, Microsoft Teams chats and meetings, or external business systems connected through Copilot Connectors. Turn on web search only when you want the agent to research beyond your organization, like competitor or customer research. Add Skills to teach the agent repeatable, multi-step work — in this video, a weekly status update email built from a real template, matching its tone, sections, and formatting automatically. Preview and test the agent before publishing, then share it with people or groups and set whether they can use it or edit it. Once your agent is live, the Monitor tab tracks total sessions, daily active users, average messages per session, average duration, and which knowledge sources get used most. For agents with strong adoption, publish them to the Built by your org section of the Agent Store for wider use across your organization. Try Copilot Agent Builder for yourself — it's included with Copilot. Subscribe to Microsoft Mechanics for more videos like this.  ► Unfamiliar with Microsoft Mechanics? Microsoft's Official Video Series for IT — Subscribe https://www.youtube.com/c/MicrosoftMechanicsSeries — Microsoft Tech Community: https://techcommunity.microsoft.com/t5/microsoft-mechanics-blog/bg-p/MicrosoftMechanicsBlog — Podcast: https://microsoftmechanics.libsyn.com/podcast ►Join us on social: — https://twitter.com/MSFTMechanics — https://www.linkedin.com/company/microsoft-mechanics/ — https://www.instagram.com/msftmechanics/ — https://www.tiktok.com/@msftmechanics  

Govern every AI agent running across your organization with Agent 365. Track agents from Microsoft, Amazon, Google, and Salesforce in one registry, sync in agents you're already running on AWS Bedrock, Google Cloud, Databricks Genie, and Anthropic Claude, and lock down shadow AI with default blocks and Execution Container isolation. Agent 365 inspects your environment for every agent Microsoft and partners have registered, then layers control on top. Reusable security policy templates apply Conditional Access, Access Packages, and Custom Security Attributes the moment you approve an agent, Tools governance blocks risky MCP servers and connectors org wide, and the Adoption Dashboard breaks down usage by group, job function, and license type for every manager in your org.  Jeremy Chapman, Microsoft 365 Director, shares how to bring every agent in your organization under one governed registry, and how to shut down shadow AI before it ever compromises your environment.  👥 Who it's for: IT admins, security engineers, and identity teams governing AI agents across Microsoft 365, Entra, Purview, Defender, and Intune; solution architects designing agent access, tools, and security policy at scale; developers and DevOps teams building and registering agents and custom MCP servers in Copilot Studio, Agent Builder, and Foundry; and team leads, finance, and IT leaders tracking agent adoption, usage, and spend across the organization. ⏱️ Chapters:  00:00 Why you need an AI agent control plane 02:00 See every AI agent across your environment with the unified agent registry 02:50 Track local AI activity with OpenTelemetry and the Agent Map 03:16 Sync AI agents from AWS Bedrock, Google Cloud, Databricks and Claude with Registry Sync 03:55 Approve, publish and block AI agents across your organization 04:42 Manage MCP servers, plugins and connectors with Tools governance 05:18 Apply reusable security policy templates across Entra, Purview, Defender and Intune 06:06 Block shadow AI and isolate local agents with Microsoft Execution Containers 06:23 Track agent adoption and usage with the Agent 365 dashboard 07:06 Set spending limits and cost alerts for AI agents 08:02 Monitor AI agent costs and buy prepaid credits 09:07 Get started with Agent 365 today  Agent 365 gives you one place to govern every AI agent running across your organization, no matter where it was built or who owns it. In this demo, the unified agent registry surfaces agents from Microsoft Foundry, Copilot Studio, Agent Builder, and SharePoint, alongside agents from Amazon, Google, and Salesforce, each with its identity, owner, permissions, and usage history attached. Local AI running on managed devices shows up too, logged through standardized OpenTelemetry and mapped visually in the Agent Map so you can trace connections between agents, people, data, and tools. Registry Sync pulls in agents you're already running on AWS Bedrock, Google Cloud, Databricks Genie, and Anthropic Claude with nothing more than a connection string, and from the same registry you approve agents for broader use, publish or pin them to specific groups, automate ownership handoffs when an employee leaves, and block any agent you don't want in production.  From there, Agent 365 puts you in control of what agents can touch and how they're secured. The Tools view gives you one place to see every MCP server, plugin, and connector your agents rely on, block the risky ones, and approve or reject new tool requests as they come in. Reusable security policy templates apply Conditional Access, Access Packages, and Custom Security Attributes the moment you approve an agent, drawing on native signals from Microsoft Entra, Purview, Defender, and Intune so your identity, data security, SecOps, and device teams enforce the same controls. Shadow AI running as unsanctioned local agents gets blocked by default through Intune policy and isolated from user sessions with Microsoft Execution Containers. Team leads and managers track adoption and usage in the Agent 365 Dashboard, filtered by group, organization, job function, or license type. IT and finance teams set monthly spending limits, per user budget caps, and weekly usage alerts for Copilot Cowork and the Work IQ API, and can pre purchase discounted credits to keep spend predictable.  🔗 Related links: ► Get started at https://aka.ms/agent365 ► Unfamiliar with Microsoft Mechanics? Microsoft's Official Video Series for IT — Subscribe https://www.youtube.com/c/MicrosoftMechanicsSeries — Microsoft Tech Community: https://techcommunity.microsoft.com/t5/microsoft-mechanics-blog/bg-p/MicrosoftMechanicsBlog — Podcast: https://microsoftmechanics.libsyn.com/podcast ►Join us on social: — https://twitter.com/MSFTMechanics   — https://www.linkedin.com/company/microsoft-mechanics/ — https://www.instagram.com/msftmechanics/ — https://www.tiktok.com/@msftmechanics  

An AI agent works inside your Windows 365 Cloud PC from a Teams chat on your phone — finding downloaded files, editing documents, and drafting emails with your laptop lid closed. Microsoft Scout, a new Autopilot agent powered by the open source OpenClaw project, runs on your own Cloud PC using your identity, your files, and your apps. Developers get a Developer Optimized Cloud PC preloaded with GitHub, VS Code, and Node.js that runs Foundry Local for token-free AI coding, and input and output protection capabilities black out the screen when anyone tries to capture a sensitive file. Bhavya Chopra, Windows 365 Partner Director, joins Jeremy Chapman, Microsoft 365 Director, to show how Windows 365 now covers every seat — end users, developers, admins, and the AI agents working alongside them.  👥 Who it's for: architects and IT teams deploying AI agents like Scout and Windows 365 for Agents, developers who want a pre-configured cloud coding environment with local AI, security and compliance teams enforcing data loss prevention on remote desktops, and admins managing Cloud PC connection health. ⏱️ Chapters: 0:00 What's new across Windows 365 Cloud PCs 0:59 Why Cloud PC adoption keeps accelerating 2:04 Healthcare and telecom Windows 365 case studies 3:14 Sign in to a full Windows desktop anywhere 4:10 Block screenshots with input and output protection capabilities 4:35 Pull table data from images with Click to Do 5:19 Resume your Cloud PC session on an iPad 5:52 Publish Cloud Apps with Windows 365 Flex 7:01 Give AI agents a dedicated Cloud PC 7:27 Run Microsoft Scout from a Teams chat 9:43 Code on a Developer Optimized Cloud PC 10:11 Run local AI with Foundry Local and Phi-4 11:57 Track Cloud PC connection health in Intune 13:03 Troubleshoot connection issues with an AI agent 14:03 What's next for Windows 365 Microsoft Scout takes instructions from a Teams chat and carries them out inside your Cloud PC — locating a report in the Downloads folder that never synced to OneDrive, summarizing it, appending the summary to the file, saving it, then drafting an email with the document attached and holding it for review. It reports progress back in Teams the whole time, so nothing requires touching the physical device. For agent workloads that need isolation instead, Windows 365 for Agents provisions a dedicated Cloud PC with its own unique identity and scoped access, and the same agentic infrastructure powers Computer Use in Researcher. Developer Optimized Cloud PCs skip the environment build entirely: sign in to a machine already loaded with GitHub, VS Code, and Node.js, then run Foundry Local against the Phi-4-mini small language model to generate an app and run it — CPU spikes visible in Task Manager and Windows Machine Learning profiling in VS Code confirm inference is local, with no cloud tokens spent to code or to run the finished app. Input and output protection capabilities black out the Cloud PC screen the instant someone attempts a screenshot from the accessing device, and extends the same block to third-party screen recorders and key loggers. Alongside it, AI-enabled Cloud PCs running at least 8 vCPUs add Click to Do, which lifts text out of non-selectable images and graphical tables straight into Excel, plus Semantic Search across the desktop. Sessions roam: disconnect on a laptop, reconnect on an iPad, and apps and display settings come back as you left them. Windows 365 Cloud Apps with Windows 365 Flex publishes individual apps from a shared Cloud PC pool with user experience sync across Windows, macOS, iOS, Android, and the browser. And in the Intune admin center, new monitoring dashboards track connection health, failures, round trip times, and license utilization in near real time, while an admin agent in preview triages issues to a root cause — like a regional cohort set to TCP-only transport instead of UDP with RDP Shortpath.  🔗 Related links:  ► Check out https://aka.ms/windows365 ► See the latest Windows 365 updates at https://aka.ms/Windows365Turns5 ► Unfamiliar with Microsoft Mechanics? Microsoft's Official Video Series for IT - Subscribe https://www.youtube.com/c/MicrosoftMechanicsSeries - Microsoft Tech Community: https://techcommunity.microsoft.com/t5/microsoft-mechanics-blog/bg-p/MicrosoftMechanicsBlog - Podcast: https://microsoftmechanics.libsyn.com/podcast ► Join us on social: - https://twitter.com/MSFTMechanics - https://www.linkedin.com/company/microsoft-mechanics/ - https://www.instagram.com/msftmechanics/ - https://www.tiktok.com/@msftmechanics