Beyond Patching: Why Vulnerability Management Matters More in the AI Era

Modern devices, seamless connectivity, cloud collaboration, and AI are transforming how employees work. But how securely are we managing the technology underneath?

Listen to a short audio summary (6 min)

Prefer reading? The full article continues below.

Modernization Needs a Strong Security Foundation

Over the years, I've seen organizations invest significantly in improving the Digital Workplace.

Better devices, modern endpoint management, seamless connectivity, cloud collaboration, and now AI-powered productivity are changing how employees work. These investments help organizations simplify operations, improve employee experiences, and enable greater flexibility.

But there's another side of modernization that deserves equal attention.

How effectively are we protecting the devices, applications, identities, and infrastructure supporting these experiences?

A modern device running vulnerable software remains a security risk. A cloud-connected application with an unaddressed weakness can expose sensitive information. And an organization delivering excellent employee experiences can still face serious disruption if vulnerabilities are not managed effectively.

We continue to see cybersecurity incidents affecting organizations across industries and geographies. They remind us that security weaknesses can have consequences far beyond IT, affecting business operations, customer trust, and organizational resilience.

This is why I believe vulnerability management should be treated as an essential part of Digital Workplace modernization, rather than simply another security operations activity.

1. AI Is Changing Vulnerability Discovery

Microsoft's November 5, 2026 announcement highlights how AI is changing vulnerability management within Windows engineering and servicing.

Microsoft is using AI-assisted capabilities, including MDASH, to identify potential vulnerabilities earlier, analyze their impact, support remediation development, and strengthen testing and validation. Human engineers remain responsible for reviewing findings and making security decisions.

MDASH is also documented as a preview capability within Microsoft Security Exposure Management, extending its relevance beyond Microsoft's internal engineering processes.

The October 2026 Windows security update addressed more than 1,100 CVEs. Microsoft explains that the higher number reflects expanded vulnerability discovery and remediation, not necessarily declining product security.

But another finding caught my attention.

Microsoft's August 2026 diagnostic data showed that, in a measured enterprise sample, fewer than half of Windows devices had installed the security update after two weeks. Approximately one in four remained unpatched after four weeks.

As AI accelerates vulnerability discovery and potentially exploitation, this gap between available protection and deployed protection becomes increasingly important.

Discovering vulnerabilities faster is valuable. Protecting enterprise environments faster is equally important.

2. Vulnerability Management Goes Beyond Patching

Patching is an important security control, but it isn't the entire vulnerability management lifecycle.

Organizations need continuous visibility into their technology landscape, including endpoints, servers, cloud workloads, applications, and exposed services.

They must understand which vulnerabilities are actively exploited, which systems are business-critical, and what immediate action is necessary.

Sometimes the right response is a security update. In other situations, temporary mitigations such as restricting access, disabling vulnerable services, changing configurations, isolating affected systems, or strengthening monitoring may be necessary.

And remediation doesn't end when a change is deployed.

Teams must verify that the vulnerability has actually been addressed and continue monitoring for unresolved exposure.

The objective is not simply to reduce the number of open vulnerabilities. It's to reduce the organization's actual security risk.

3. The Bigger Challenge: Deploying Protection at Enterprise Scale

Consider a large organization managing thousands of endpoints across countries, business units, hardware models, and applications.

A critical vulnerability is identified, and a fix becomes available.

Deploying it immediately across the entire environment may introduce compatibility or operational risks. Delaying it without appropriate protection could create an equally serious security exposure.

Microsoft also warns that once security updates become public, attackers can analyze the fixes and potentially develop exploits within hours. That makes timely risk assessment and remediation increasingly important.

Security and infrastructure teams need to assess the urgency, identify the most exposed systems, implement immediate mitigations where required, and validate the permanent fix against representative devices and business applications.

Controlled deployment rings, application compatibility testing, rollback readiness, and post-deployment monitoring remain essential. For actively exploited vulnerabilities, emergency change procedures and accelerated deployment may be necessary.

Modern capabilities such as Microsoft Intune and Windows Autopatch can help streamline deployment. I recently explored the additional deployment controls in Windows Autopatch Adds Policy-Driven Update Approvals.

This also connects to my earlier discussion on investing in testing before production.

The real differentiator isn't the tool alone. It's the ability of security, endpoint engineering, application owners, and service management teams to respond quickly while maintaining business continuity.

4. How Today's Security Tools and AI Can Help

Organizations already invest in vulnerability management platforms such as Microsoft Defender Vulnerability Management, Qualys, Tenable, Rapid7, and other security solutions.

These tools help discover vulnerabilities, identify affected assets, prioritize remediation, and monitor security exposure.

Microsoft Security Exposure Management adds broader risk context, while Security Copilot introduces AI-assisted investigation and analysis.

Extended Detection and Response (XDR) complements these capabilities by correlating security signals across endpoints, identities, email, and other workloads to help security teams detect, investigate, and respond to potential attacks.

Another relevant development is Microsoft's Vulnerability Remediation Agent in Intune, currently in public preview.

The agent uses Defender Vulnerability Management data to identify and prioritize vulnerabilities and provide step-by-step remediation guidance.

It does not automatically patch devices. Administrators remain responsible for implementing changes, and the agent has specific licensing, platform, and operational limitations.

Microsoft's November announcement focuses on Windows security engineering and servicing. These enterprise security tools are complementary capabilities rather than a single workflow introduced in that announcement.

For organizations, the opportunity is to connect vulnerability intelligence, security operations, endpoint management, and business risk decisions more effectively.

Having multiple tools is useful. Having a coordinated operating model that turns their findings into timely protection is what matters.

Beyond patch compliance: Moving towards risk-based vulnerability management, AI-assisted discovery, controlled remediation, and continuous protection.

5. Making Security Part of Modernization

As organizations continue modernizing their Digital Workplace, I believe a few priorities deserve greater attention.

• Build security into modernization: Every new device, application, cloud service, and AI capability needs clear ownership, security baselines, and lifecycle management.
• Prioritize real exposure: Focus on exploit activity, business-critical assets, and potential impact rather than vulnerability counts alone.
• Strengthen testing and deployment: Establish representative pilot groups, emergency remediation processes, recovery plans, and clear accountability.
• Measure effective protection: Track time to remediation, critical asset exposure, failed deployments, and unresolved exceptions.
• Stay continuously vigilant: Vulnerability management should operate throughout the technology lifecycle, not only during monthly patching or audit reviews.

These practices help organizations balance security, operational stability, and employee productivity.

Final Thoughts

The Digital Workplace is becoming more intelligent, connected, and employee-centric.

Organizations are investing in modern devices, cloud collaboration, automation, and AI to create better experiences and improve productivity.

But beneath those investments sits an equally important responsibility: protecting the technology and information employees depend on.

Microsoft's latest announcement demonstrates how AI can accelerate vulnerability discovery and remediation. Yet the broader enterprise challenge remains understanding exposure, implementing the right controls, validating changes, and delivering protection safely at scale.

A truly modern Digital Workplace isn't just one that helps employees work better. It's one that enables them to work securely, reliably, and with confidence.

And that is where effective vulnerability management becomes a key differentiator.

References and Further Reading

• Microsoft — A New Standard for Vulnerability Management in the AI Era
• Microsoft — MDASH Overview (Preview)
• Microsoft — Defender Vulnerability Management
• Microsoft — Security Exposure Management
• Microsoft — Vulnerability Remediation Agent in Intune
• CISA — Known Exploited Vulnerabilities Catalog

Starting Your AI Journey: A Curated Microsoft AI Learning Navigator

Not sure where to begin with AI? Here's a practical collection of learning resources, from fundamentals and Copilot to AI agents and enterprise adoption.

Watch: Your AI learning starting point in 60 seconds

Over the past few months, I've come across a familiar question from people who are beginning to explore AI: “With so many learning resources available, where should I start?”

It's a reasonable question. AI is evolving rapidly, and while there is no shortage of learning materials, finding the right starting point isn't always easy.

There are courses, videos, tutorials, documentation, community projects, and certifications everywhere. For someone just starting, choosing what to learn first can be more confusing than learning the technology itself.

I've been exploring AI alongside my work in enterprise technology, and I continue to discover something new almost every day.

So I thought I'd put together a small collection of useful learning resources, predominantly from Microsoft's ecosystem, for anyone looking to begin or gradually deepen their AI knowledge.

This isn't an exhaustive directory or an official learning curriculum. It's simply a curated starting point, with references to resources that I believe are worth exploring.

1. Completely New to AI? Start Here

Microsoft AI Skills Navigator

If someone asked me to recommend just one starting point, this would be it.

Microsoft's AI Skills Navigator brings together learning paths, videos, training materials, and credentials. You can explore resources based on your interests, role, and learning goals.

Whether you're a student, business professional, IT engineer, developer, or technology leader, it's a useful place to begin.

Microsoft Learn – AI Fluency

This learning path introduces AI fundamentals, generative AI, prompting, responsible AI, and Microsoft Copilot.

My suggestion? Start with the fundamentals. Understand what AI can do, where it struggles, and why human judgment still matters.

Don't rush into building agents before understanding the basics.

If you prefer live learning alongside self-paced courses, you can also explore Microsoft Discovery Sessions for expert-led discussions, demonstrations, and practical insights into Microsoft technologies.

2. Want to Use AI in Your Everyday Work?

Once you're comfortable with the terminology, start experimenting.

Get Started with Microsoft Copilot

A useful introduction to interacting with Copilot, writing effective prompts, and understanding its capabilities.

Microsoft Copilot

You can practice with simple activities: summarize an article, draft an email, compare two ideas, or create a learning plan.

If your organization uses Microsoft 365 Copilot, explore how AI can support your work across Outlook, Teams, Word, Excel, and PowerPoint.

Start with one activity you already perform regularly.

For example, take a lengthy meeting summary and ask Copilot to identify decisions, action items, and unresolved questions.

Then check whether the output is correct.

That's an important part of learning AI: knowing how to question and validate what it produces.

3. Interested in Building AI Agents?

This is where learning becomes particularly interesting.

Microsoft Copilot Studio

Copilot Studio helps you understand how agents can use knowledge, instructions, tools, and workflows to support business activities.

Copilot Studio Agent Academy

This is one of the practical learning resources I'd recommend exploring.

Agent Academy offers two learning tracks: Recruit for foundational skills and Operative for more advanced agent development. Both tracks support the Standard harness and Microsoft Copilot harness, with hands-on exercises and completion badges. The Commander track is listed as coming soon.

Through these tracks, you can explore how to build agents, connect knowledge sources, work with tools, and introduce automation.

You don't need to build a sophisticated solution immediately. Start with one small scenario and understand how the different components work together.

4. For Developers and Technical Professionals

If you want to understand what happens behind an AI application, Microsoft's open-source learning projects are worth exploring.

Generative AI for Beginners – GitHub

A 21-lesson course covering language models, prompt engineering, responsible AI, RAG, application development, and more.

AI Agents for Beginners – GitHub

An 18-lesson course exploring agent design patterns, tool usage, multi-agent systems, agentic RAG, and production considerations.

AI for Beginners – GitHub

For those interested in traditional AI and machine learning fundamentals, including neural networks and computer vision.

AI Applications and Agents on Azure – Microsoft Learn

A learning path introducing Microsoft Foundry, generative AI, agents, language processing, and other AI workloads.

These are particularly useful for developers, architects, and engineers looking to understand AI beyond the user interface.

Some exercises require programming knowledge, Azure access, or additional services.

5. For IT Leaders: Learn How Organizations Adopt AI

Learning AI isn't only about writing prompts or building agents.

For organizations, the bigger questions involve adoption, security, data protection, governance, cost, and operational ownership.

Microsoft 365 Copilot Adoption Resources

Explore adoption guidance, business scenarios, user enablement, and resources for introducing Copilot into an organization.

Advanced Copilot Adoption Resources

Useful for technology leaders, adoption managers, and IT administrators planning broader deployments.

Microsoft Responsible AI

Understand the principles and considerations behind responsible AI development and deployment.

These topics matter because a successful AI initiative needs more than a working demonstration.

Someone must think about who owns the solution, what information it can access, how its performance is monitored, and what happens when something goes wrong.

6. For Organizations: Create Your Own AI Learning Plan

Microsoft Learn also offers a useful way for organizations to structure AI learning across their teams.

Microsoft Learn Plans allows team leaders to organize learning resources into milestones, invite employees, and track their progress.

For example, an organization could create a four-week AI learning plan covering AI fundamentals, responsible AI, Microsoft Copilot, and introductory AI agents.

Employees can earn achievements through eligible learning activities, while formal Microsoft credentials require separate assessments.

I find this particularly useful for organizations looking to encourage structured AI upskilling rather than simply sharing individual training links.

Learn how to create and manage Microsoft Learn Plans

7. Keep AI Atlas Alongside Your Learning

While exploring these resources, you'll encounter terms such as LLMs, RAG, grounding, orchestration, AI agents, MCP, and responsible AI.

I've been building AI Atlas as a complementary learning and reference resource.

It brings together AI vocabulary, AI ecosystems, Microsoft Copilot concepts, Digital Workplace technologies, responsible AI, and enterprise adoption topics.

For example, if you're following an Agent Academy exercise and come across an unfamiliar concept, you can explore the terminology in AI Atlas before continuing.

The idea isn't to replace Microsoft's official training. It's to make the broader AI landscape easier to understand and connect.

8. How Would I Approach the First 30 Days?

I wouldn't recommend spending the first month simply watching videos.

I'd combine learning with practice.

Week	What I'd focus on
Week 1	Explore AI Skills Navigator and complete introductory AI lessons.
Week 2	Practise prompting and use AI for a few everyday tasks.
Week 3	Explore Copilot Studio, Agent Academy, or a developer course based on your interests.
Week 4	Try one small use case, evaluate the results, and identify your next learning goal.

You won't become an AI expert in 30 days. That's not the objective.

The goal is to become comfortable enough to continue learning independently.

Final Thoughts

If there's one thing I've learned from working across different generations of enterprise technology, it's that we never really finish learning.

AI is no different.

You don't need to understand every model, platform, or framework. You don't need to become a developer unless that's where your interests take you.

Start with the fundamentals. Experiment with something relevant to your work. Learn from the mistakes. Gradually explore more advanced concepts.

For organizations, the same principle applies. Encouraging continuous learning and giving employees a structured path can be just as important as investing in AI technology itself.

Although I've focused on Microsoft resources here, AI learning shouldn't be limited to one vendor. Many underlying concepts apply across ChatGPT, Gemini, Claude, and other platforms.

I'll continue refreshing this collection as useful resources emerge.

The best way to begin your AI journey is not to wait until you find the perfect learning path. It's to choose one and start.

Author's note: This is an independent collection of learning resources, not an official Microsoft curriculum or endorsement. Most learning content is freely accessible, although some labs, cloud services, product capabilities, and certifications may require licenses or additional costs. Resources and prerequisites can change over time.

Copilot Projects, Workspaces and Agents: What Should You Use and When?

Microsoft 365 Copilot experiences for organizing work, collaborating, and using AI assistance.

Microsoft Copilot is evolving beyond a conversational assistant into a broader environment for organizing work, collaborating with teams, and using specialized AI capabilities.

With Microsoft's recent evolution of Copilot Notebooks into Workspaces and the introduction of Projects, I've been looking more closely at how these experiences fit together. While the capabilities are interesting individually, the more practical question is which one an everyday employee should use, and when.

Understanding the Differences

Copilot offers several complementary experiences, each designed for a different way of working. The following visual summarizes their primary purposes and where they fit into everyday productivity.

Microsoft 365 Copilot experiences at a glance — complementary capabilities, not a required sequence.

What Happened to Copilot Notebooks?

Microsoft is evolving Copilot Notebooks into Copilot Workspaces, bringing a stronger focus on shared context and team collaboration.

Existing notebook content and sharing arrangements carry forward. Importantly, a private notebook does not automatically become accessible to everyone.

The rollout is phased, beginning in late October 2026, so availability may vary across organizations.

Copilot Projects remains a separate experience, primarily intended for organizing individual work.

A Simple Workplace Example

Consider an employee preparing a quarterly service review.

They might use Copilot Chat to summarize information, Projects to organize personal analysis and working notes, Pages to refine the report, and Workspaces to share review materials and collaborate with colleagues.

If preparing these reviews becomes a recurring organizational requirement, a Copilot Agent could provide standardized assistance using approved knowledge and instructions.

This doesn't mean Projects automatically become Agents or that every task requires an Agent.

One Governance Consideration

As employees increasingly use AI to organize and perform work, organizations should consider what happens to the information and capabilities they create.

Who maintains an important Workspace when its creator changes roles? Where should business-critical outputs be retained? Who owns and reviews an Agent used by multiple employees?

The answers depend on the experience, its storage and sharing model, and organizational policies.

This connects to a broader topic I recently explored: When an AI Agent Outlives Its Owner: Digital Succession and Supplier Dependency.

Final Thoughts

Copilot Chat helps with everyday assistance. Projects organizes personal work. Pages preserves useful outputs. Workspaces supports collaboration. Agents provide specialized capabilities.

The value lies not in using every tool, but in choosing the right experience for the work—and managing what becomes important to the organization.

Official Microsoft References

• Copilot Notebooks is now Copilot Workspaces
• Compare Microsoft Loop, Copilot Pages and Notebooks
• Microsoft Copilot Agents documentation

When an AI Agent Outlives Its Owner: Digital Succession and Supplier Dependency

AI agents may outlive their creators. Ownership, knowledge, and accountability must endure.

Listen to a short audio summary (6 min)

Prefer reading? The full article continues below.

What happens when the person who built an AI agent is no longer available to support it?

Imagine an organization deploying an AI agent to automate SharePoint permission attestations.

The agent coordinates access reviews, follows up on pending approvals, and generates compliance reports. Everything works as expected, and over time, the organization starts depending on it.

Then the engineer who developed it leaves. Perhaps the engineer is an employee, or perhaps they work for a supplier whose contract is ending.

The agent continues to exist. But who understands its architecture, integrations, business rules, and recovery procedures?

Can another engineer take over without depending on the original developer?

This is where AI agent ownership becomes a digital succession challenge.

In my earlier article, Who Owns an AI Agent After It Goes Live?, I explored accountability after deployment.

Here, I want to look at something closely related but different: how organizations retain the knowledge and operational control behind the AI assets they create.

The offboarding risk we may be overlooking

Employee offboarding is familiar territory for most IT organizations. Accounts are disabled, access is revoked, and responsibilities are transferred.

But AI agents introduce dependencies that may not be immediately visible.

An agent could rely on Power Automate workflows, APIs, connectors, knowledge sources, service identities, and configuration decisions made by one engineer.

Microsoft already documents how Copilot Studio agents can become orphaned when owners leave, and how Power Automate flows may fail when they depend on connections associated with departed employees.

Microsoft's agent registry can help identify agents without owners, while Agent Builder supports multiple owners to reduce reliance on a single individual.

These are useful capabilities. But identifying an orphaned agent or assigning a new owner is only part of the process.

Transferring ownership does not automatically transfer operational knowledge.

There are security considerations too. Depending on assigned privileges, a new owner may gain access to existing agent transcripts. That makes permissions and data-access reviews an important part of succession.

The AI Asset Passport: Keeping knowledge within the organization

I believe every production AI agent should have a simple, maintained record of what it does, how it works, and who can support it.

I would call this an AI Asset Passport.

Document the agent. Preserve the knowledge. Protect continuity.

It is a proposed governance practice, not an official Microsoft framework.

Area	What should be captured
Identity	Purpose, environment, business criticality
Ownership	Business owner, technical custodian, backup owner, supplier contacts
Dependencies	Workflows, APIs, connectors, knowledge sources
Security	Identities, permissions, data sensitivity
Operations	Monitoring, troubleshooting, recovery procedures
Knowledge	Architecture, source repositories, runbooks, change history
Succession	Successor details, handover evidence, validation date

The passport doesn't need to become another lengthy document.

It could be a lightweight record linking to existing CMDB entries, SharePoint repositories, source-control systems, and operational documentation.

What matters is that these records remain accessible to authorized teams, version-controlled where appropriate, and regularly updated.

Knowledge retention cannot be a one-time activity performed during deployment or offboarding.

Six practical controls for digital succession

I would build succession into existing enterprise service management and AI governance practices through six controls:

1. Accountable ownership: Assign a business owner, technical custodian, and successor.

2. AI Asset Passport: Maintain an inventory of purpose, dependencies, ownership, and operational references.

3. Knowledge continuity: Preserve architecture, configurations, runbooks, and change history in enterprise-controlled repositories.

4. Supplier succession: Establish backup resources, knowledge-transfer obligations, and exit procedures.

5. Operational validation: Verify that another authorized engineer can troubleshoot, modify, and recover the agent.

6. Continuous assurance: Periodically review ownership, documentation freshness, dependencies, and succession readiness.

These controls are not intended to introduce bureaucracy. They are practical measures to prevent operational knowledge from disappearing when people change.

This approach also aligns with NIST's AI Risk Management Framework, which addresses AI system inventories (Govern 1.6), defined roles and responsibilities (Govern 2.1), safe decommissioning (Govern 1.7), and third-party risk management (Govern 6).

A practical example: SharePoint access attestation

Consider an organization that engages a supplier to develop an AI-assisted SharePoint permission attestation solution.

The solution uses an agent to coordinate reviews, Power Automate to orchestrate workflows, and approved Microsoft Graph integrations to retrieve relevant information.

A supplier engineer develops and supports the solution.

A few months later, that engineer is reassigned to another project.

The organization may have the agent, workflows, and necessary licenses, but the detailed knowledge of exception handling, approval logic, or technical dependencies may sit primarily with that engineer.

This does not mean the agent automatically stops working. The risk is that troubleshooting, maintenance, or future changes become difficult.

With a succession plan, the transition looks different.

The replacement engineer can access the AI Asset Passport, review current architecture and runbooks, validate workflow connections, and execute a controlled attestation test.

The service owner confirms that approvals, exceptions, reporting, and monitoring work before accepting the handover.

The organization has not simply reassigned ownership. It has demonstrated continuity.

Supplier dependency: The challenge deserves more attention

This is where I see a broader operational concern.

Across enterprise IT, knowledge transfer has always been a challenge, particularly in supplier-managed environments.

From my experience in enterprise service ownership and supplier management, I've seen how resource transitions, knowledge gaps, and unclear accountability can introduce operational risks, even when the underlying technology is working well. As AI agents become part of business operations, these challenges deserve attention from the outset.

People move between projects, contracts change, critical resources resign, and sometimes supplier teams are reorganized with very little notice. Often, the gaps become visible only when an incident occurs or a replacement engineer needs to make a change.

AI agents could make this challenge more significant.

A supplier may develop multiple agents across HR, Finance, Procurement, and IT. Each agent may contain business rules, prompts, workflows, integrations, and configuration decisions that are not obvious from the user interface.

If that knowledge stays with a few supplier engineers, the organization becomes dependent on individuals rather than on a sustainable operating model.

Suppliers may change. Enterprise knowledge and control must remain.

I believe supplier AI exit readiness should become part of vendor governance from the beginning of an engagement.

That means establishing a few expectations:

• Enterprise-controlled knowledge: Architecture, configuration references, runbooks, test evidence, and change records should be maintained in approved central repositories, accessible to authorized customer teams.
• Continuous knowledge transfer: Suppliers should update documentation as solutions evolve, with periodic walkthroughs and reviews rather than a last-minute handover.
• Resource succession: Critical agents should have identified primary and backup support resources, alongside an accountable internal custodian.
• Transition readiness: Changes in supplier personnel should trigger dependency checks, knowledge validation, and formal service-owner acceptance.
• Contractual clarity: Deliverables, intellectual-property rights, transition assistance, and access revocation should be addressed in supplier agreements.

For proprietary solutions, organizations may not always receive source code or full control over the underlying technology. In those cases, data portability, configuration exports, and documented transition procedures become important.

The goal is not to eliminate suppliers or reduce their contribution.

It is to ensure that the knowledge needed to operate enterprise AI assets remains available to the organization, regardless of who delivers the service.

Measuring whether succession really works

I would keep the measurements simple.

Are production agents assigned accountable owners and successors? Is their documentation current? Have supplier-managed agents completed knowledge-transfer reviews?

And perhaps most importantly, how long would it take an authorized replacement engineer to take over and demonstrate that the agent can be operated safely?

For business-critical agents, I would introduce a periodic succession test:

Can someone other than the original creator support this agent using the knowledge, permissions, and tools the organization has retained?

True ownership is proven when the next person can operate independently.

If not, the handover is not complete, regardless of how many documents have been signed off.

From Governance Principles to Practical Action

Understanding digital succession is only the beginning. Organizations also need repeatable processes to manage ownership, access, knowledge transfer, operational continuity, and lifecycle changes.

To help translate these principles into practice, I've developed the Digital Workplace & AI Lifecycle Toolkit—an interactive resource covering AI agent governance, employee onboarding and off-boarding, Microsoft 365 services, supplier transitions, device lifecycle, and continuous improvement.

Access Toolkit here: Explore the Interactive Toolkit in AI Atlas

From AI ownership to long-term stewardship

Digital succession should become part of existing offboarding, IAM, change management, service ownership, and supplier governance processes.

As organizations deploy more agents, they will inevitably face employee exits, supplier transitions, platform changes, and changing business requirements.

We should be thinking about these situations while designing and deploying agents, not after losing the people who understand them.

The number of agents deployed may demonstrate the pace of AI adoption.

But the ability to maintain those agents independently, securely, and consistently will demonstrate operational maturity.

An AI agent should never become an operational mystery simply because the person who built it has moved on.

References

• Microsoft — Reassign ownership of orphaned Copilot Studio agents
• Microsoft — Manage orphaned Power Automate flows
• Microsoft — Governance and lifecycle actions for agents available in Microsoft 365 admin center
• NIST — AI Risk Management Framework

Author's note: The AI Asset Passport and six digital succession controls are proposed enterprise governance practices, not official Microsoft or NIST frameworks. The SharePoint attestation example is illustrative and does not represent a specific organization.

Three Months of Learning: 10 Perspectives on Enterprise AI, Digital Workplace & Governance

Over the past three months, I've been writing about developments across enterprise technology, from Digital Workplace and Microsoft 365 to AI adoption, service management, security, and governance.

Looking back, what started as individual learning notes has gradually become a collection of connected perspectives.

Some articles focus on emerging technologies, while others explore the practical side of running enterprise IT services: testing, observability, operational ownership, financial governance, and the challenges that come with AI adoption.

As we begin November, I thought it would be a good time to bring together ten articles from August through October 2026.

This isn't a ranking or an exhaustive list. It's a collection of topics I've enjoyed exploring, organized to help readers discover something relevant to their interests.

Watch: Three Months of Learning in 60 Seconds

The visuals in this video are AI-generated and represent a conceptual scenario, not an actual implementation or prediction.

Digital Workplace & Enterprise AI

1. The Digital Workplace Is Becoming an AI Workplace: But the Foundations Haven't Changed

The Digital Workplace is evolving into an AI-enabled environment, but the foundations remain important. This article explores how employee experience, endpoints, collaboration, identity, security, and service governance continue to shape the modern workplace.

2. Enterprise AI Is Evolving Fast: Are We Ready for a Multi-AI Workplace?

With Microsoft Copilot, ChatGPT, Claude, Gemini, and other AI platforms becoming part of the enterprise conversation, organizations have more choices than ever. I explore what this means for adoption, integration, and governance.

Service Operations, Testing & Observability

3. AI-Enabled Service Management: Automating Work, Evolving People's Roles

AI and automation are changing how IT services are delivered. This article looks at the opportunities to improve service operations while maintaining clear ownership, accountability, and human judgment.

4. Is Your Organization Investing Enough in Testing Before Production?

Technology is evolving quickly, but are our testing environments and validation practices keeping pace? I share why testing, change governance, phased deployments, and operational readiness deserve continued investment.

5. Observability Is Taking Center Stage: What AI-Driven Remediation Changes, and What It Doesn't

As enterprise environments become more interconnected, understanding service health and employee experience is increasingly important. This article explores observability, AI-assisted investigation, automated remediation, and why operational ownership still matters.

AI Economics & Agent Governance

6. FinOps for AI: Understanding Tokens, Copilot Credits and the Real Cost of AI

AI adoption brings new consumption models and financial considerations. From tokens to Copilot credits, I explore how organizations can better understand usage, manage costs, and connect AI investments to business value.

7. Building an Agentic Center of Excellence: Do We Really Need Another CoE?

As organizations introduce AI agents, governance structures are evolving too. Rather than immediately creating another CoE, this article explores how existing enterprise capabilities might be extended to support agentic AI.

8. Who Owns an AI Agent After It Goes Live? Building the Enterprise Agent Operating Model

Deploying an AI agent is only the beginning. Who owns it when something goes wrong? This article explores operational accountability, agent identities, access controls, monitoring, and lifecycle management.

Security, Regulation & Responsible AI

9. The EU AI Act Enters Enforcement: What It Really Means for Microsoft 365 Copilot

August 2026 marked an important milestone for the EU AI Act, with new transparency obligations and enforcement powers for applicable provisions. This article explores what those developments mean for enterprise AI adoption, why Copilot use cases need to be assessed individually, and how organizations can prepare for requirements taking effect in later phases.

Looking Ahead: The Enterprise of 2035

10. The Enterprise of 2035: Who Will Lead the AI Workforce?

I wanted to close this collection with a different perspective, one that looks beyond today's technology announcements.

What might the enterprise look like in 2035, when people and AI agents could be working alongside each other at a much greater scale?

This article explores ten possibilities for the future workplace, including AI agent identities, evolving responsibilities, human leadership, and the importance of developing judgment and experience in an AI-enabled world.

Looking Forward to November

October brought plenty of interesting developments, and November looks promising too.

Microsoft Ignite 2026, scheduled for November 17–20, is just around the corner. I'm particularly interested in what Microsoft shares across Copilot, AI agents, security, governance, and the evolving Digital Workplace.

There will be more to explore, but as always, I prefer to look beyond the announcements and understand what they mean for enterprise adoption and day-to-day operations.

I'll be following the developments and sharing my learning notes and perspectives here on Exchange Anywhere.

Explore More with AI Atlas

If you're exploring some of the terminology and technologies discussed in these articles, you can also visit my AI Atlas.

It brings together AI vocabulary, technology landscapes, enterprise adoption frameworks, governance concepts, and other learning resources in one place.

Closing Thoughts

Looking back at these three months, one thing stands out to me.

The technologies may be changing rapidly, but the fundamentals of enterprise IT haven't disappeared. Reliable services, secure environments, good governance, responsible investment, and clear accountability remain just as important.

AI is adding a new dimension to all of this, and there's still plenty for us to learn.

Thanks for following along, and here's to more learning and conversations in November.

Stay tuned to Exchange Anywhere for more updates.

Discover, Govern and Contain AI Agents: Microsoft's Expanding Enterprise Security Approach

Discover, Govern and Contain AI Agents: Microsoft's Expanding Enterprise Security Approach

As AI agents become part of enterprise operations, an important question emerges: How do organizations discover agents, govern their access, and respond when they behave unexpectedly?

These agents may be developed through Microsoft Copilot Studio, hosted on third-party platforms, or installed directly on employee workstations.

Microsoft's October 29 announcement on the general availability of local AI agent discovery through Defender for Endpoint marks another important step toward enterprise-wide AI visibility.

1. Discover: Where are our AI agents?

Microsoft provides different discovery mechanisms depending on where an agent operates.

• Microsoft 365 agents: The Microsoft 365 Agent Registry provides visibility into supported Microsoft, partner, organizational, and user-shared agents.
• Third-party cloud agents: Microsoft Agent 365 supports connected-platform synchronization from Amazon Bedrock, Google Vertex AI, Salesforce Agentforce, Databricks Genie, Anthropic Claude Managed Agents, Oracle Generative AI Agents, and Snowflake Cortex.
• Local AI agents: Microsoft Defender for Endpoint discovers supported agents such as Claude Code and Cursor on managed Windows and macOS devices, including their configured Model Context Protocol (MCP) servers.

For local agents, administrators can navigate to Microsoft Defender portal → Assets → AI agents → Local agents.

Discovery begins after Defender observes supported agent activity on an onboarded device. An installed but inactive agent may not appear immediately.

Microsoft's October 29 announcement describes coverage for more than 50 supported agents and MCP servers. However, no discovery mechanism guarantees visibility into every custom or unsupported agent.

2. Govern: Where does Agent 365 fit?

Microsoft Agent 365 provides a centralized governance approach, while the Microsoft 365 Agent Registry offers inventory, ownership, and administrative visibility.

Microsoft Defender contributes security discovery and investigation. Microsoft Entra and Purview provide complementary identity, access, and data security capabilities.

For connected third-party platforms, an important distinction applies:

Connections are read-only by design. Agent 365 does not invoke, execute, or modify agents in the source platform through these connections.

The exception is an administrator's explicit action to remove a synchronized agent, which requires a separate, optional deletion permission.

Organizations can withhold that permission unless there is a clear operational requirement.

Ownership matters beyond initial onboarding

Connected-platform governance also introduces operational responsibilities:

• Synchronization: Registry synchronization is manual. Changes or deletions in the source platform may not automatically appear in the registry, making periodic reconciliation important.
• Credentials: Supported connections require key-based credentials. Credential rotation, secure storage, and expiry monitoring need clear ownership.
• Administrative responsibilities: The Agent 365 AI administrator manages connection setup, validation, and synchronized-agent verification, while the source-platform administrator provisions credentials and grants permissions.

The Microsoft 365 Agent Registry also highlights agents without owners, including situations where a shared agent's creator has left the organization. This reinforces an important governance principle: agent ownership must be reviewed throughout its lifecycle, not just assigned during deployment.

These are practical examples of why AI agent governance requires shared accountability between platform, identity, and security teams.

3. Contain: When an agent behaves unexpectedly

Consider an employee running an AI coding assistant on a managed corporate laptop, configured with an MCP connection to an internal development service.

Defender discovers the supported agent, its associated device and account, and the configured MCP server.

Security teams can investigate its configuration, including whether the agent automatically approves its own actions, and assess the associated identities and potential exposure.

Depending on the deployment, typical incident-response steps may include:

• Microsoft 365 agents: Use supported administrative controls to block or remove the agent.
• Local endpoint agents: Restrict the application, investigate the endpoint, or isolate the device when warranted.
• Third-party cloud agents: Use source-platform controls to disable the agent or revoke affected credentials and integrations. Where supported and explicitly authorized, Agent 365 may provide a synchronized-agent removal action.

These are general incident-response approaches, not a single automated containment capability delivered by Defender discovery.

Discovering an agent does not automatically provide a universal kill switch.

4. Licensing, availability and limitations

Microsoft's October 29 announcement describes local AI agent discovery on Windows and macOS as generally available.

An earlier Microsoft Learn implementation page still labels macOS discovery as preview. Because these sources differ, organizations should confirm current platform-specific availability before deployment.

Other considerations include:

• Licensing: Defender for Endpoint Plan 2 supports discovery and advanced hunting. Agent risk levels, risk indicators, and security recommendations require Microsoft 365 E7, or Microsoft Agent 365 together with Defender for Endpoint Plan 2.
• Cloud availability: Local agent discovery requires Microsoft's commercial cloud; sovereign and national clouds are not supported.
• Runtime protection: Windows runtime prompt-injection protection remains in preview and is separate from discovery.
• Observability: Connected-platform observability data collection is a separate capability associated with the Frontier preview program; agent synchronization should not be confused with full runtime monitoring.

My perspective

Enterprise AI governance must extend beyond centrally managed agents.

Digital Workplace, EUC, identity, and security teams increasingly need visibility into local AI tools, third-party agents, their permissions, and the enterprise systems they can access.

But visibility alone is not enough. Organizations also need clear responsibility for agent ownership, registry synchronization, credential lifecycle, access reviews, and incident response.

Three questions should guide the operating model:

Can we discover the agent? Who governs it? How can we contain it?

This complements my earlier article, Who Owns an AI Agent After It Goes Live?, which explored ownership and operational accountability.

The next step is translating those principles into practical security operations.

Official Microsoft references

• Microsoft Defender — Local AI agent discovery announcement
• Microsoft Learn — Discover local AI agents with Defender for Endpoint
• Microsoft Learn — Connected platforms in Microsoft Agent 365
• Microsoft Learn — Agent Registry overview

Microsoft's Becoming Frontier: 8 Practical Guides for AI Adoption Across Business Functions

Microsoft has released a collection of eight practical e-books under its Becoming Frontier initiative, focused on how Microsoft 365 Copilot and AI agents can support different business functions.

What makes this collection useful is its role-based approach. Rather than presenting AI as a single technology deployment, the guides explore how different teams can apply AI to their everyday work, decisions, and business processes.

What's included?

The collection covers eight business functions:

• Leadership: Decision-making, work redesign, and scaling AI adoption.
• Marketing: Customer insights, campaign planning, and workflow improvement.
• Finance: Financial insights, controls, and accountability.
• HR: Workforce processes and preparing employees for AI-enabled work.
• IT: Modernizing operations and establishing a secure AI foundation.
• Legal: Improving legal workflows while maintaining trust and governance.
• Sales: Customer engagement, pipeline visibility, and sales decisions.
• Customer Service: Case resolution, SLA risk, escalation management, and customer outcomes.

My perspective

One of the challenges with enterprise AI adoption is moving beyond technology availability to meaningful business use cases.

An AI agent supporting IT operations will have different data, permissions, success measures, and operational responsibilities from an agent supporting Finance or Customer Service.

That's why a role-based approach matters.

Organizations need to understand where AI can create value, how employees will work alongside it, and who will maintain accountability once these capabilities become part of everyday operations.

This also connects with my earlier article, What It Takes for an Organization to Become a Frontier Firm, where I explored the broader organizational transformation needed to move beyond AI experimentation. Microsoft's role-based guides complement that perspective by providing practical starting points for applying AI across different business functions.

These guides provide a useful starting point for organizations looking to translate AI strategy into everyday work and measurable business outcomes.

Explore the guides

Microsoft has made the collection available through its official resource page:

Becoming Frontier — Microsoft 365 Copilot and Agents e-book collection

These are Microsoft learning resources. Individual guides may require registration.

Final thought: The opportunity with enterprise AI isn't simply to introduce more agents. It's to identify where they can improve work, establish the right controls, and deliver measurable business outcomes.

Copilot Cowork Skills for Leaders: A Practical Guide and a Governance Perspective

I came across a 97-slide guide, The Big Book of Cowork Prompts & Skills for Leaders & Executives, shared by Abigail Bohrer of Microsoft on LinkedIn.

It offers practical examples of how leaders can move beyond individual prompts and use reusable skills to simplify everyday work.

The guide's original publisher has not been independently verified.

What's Inside?

The guide covers:

• 8 practical prompts for calendar planning, open actions and escalation management.
• 20 reusable skills for meeting preparation, decision-making, vendor negotiations and leadership activities.
• 8 multi-skill workflows, including a weekly briefing that combines 12 skills.
• Guidance on sharing skills, managing credits and measuring productivity improvements.

What Caught My Attention?

Beyond personal productivity, reusable skills introduce some familiar enterprise considerations.

Ownership: When skills are shared across teams, who maintains, reviews and retires them?

Governance: What information can skills access, and how should sensitive employee or business data be handled?

Cost: How do organizations monitor Copilot credit consumption and ensure it delivers measurable value?

Adoption: Can improvements in meeting time, decision turnaround and follow-up activities demonstrate real productivity gains?

These questions connect with my earlier discussions on AI agent ownership and AI FinOps.

My Perspective

The interesting shift isn't simply having more prompts. It's turning useful instructions into reusable skills that support repeatable business workflows.

For enterprises, that opportunity also brings responsibility for ownership, access, lifecycle management and cost.

Productivity tools scale faster than the habits that govern them. Reusable skills are no exception.

References

• Abigail Bohrer's LinkedIn post sharing the guide
• Microsoft Adoption — Copilot Prompt Gallery

From GitHub Copilot Harness to Microsoft Copilot Harness: What's Changed?

Microsoft Copilot Harness — A Shared Execution Foundation Across Copilot Studio and Microsoft Foundry.

Back in August, I wrote about the different harnesses available in Microsoft Copilot Studio: Standard, Copilot Chat and GitHub Copilot. I also shared Microsoft's Agent Academy learning paths for getting hands-on with them.

Microsoft's October 28 announcement brings an interesting update to that earlier discussion.

The GitHub Copilot harness in Copilot Studio is now called the Microsoft Copilot harness.

But there is more to this announcement than a name change.

What Has Changed?

Microsoft is extending a shared harness foundation across its Copilot experiences, including Microsoft Foundry.

In Copilot Studio, the change is primarily about naming. Existing agents continue working without changes to their capabilities or billing, and the updated terminology will appear over the coming weeks.

In Microsoft Foundry, however, Microsoft has introduced the Copilot harness in Foundry Agent Service in public preview.

It provides a managed execution environment for more complex agents, bringing together reusable skills, connected tools, file and code execution, subagent delegation and working-context management. Developers can also manage agent definitions alongside their application code.

Separately, Microsoft announced the general availability of long-running resilience for hosted agents, helping workflows recover from interruptions. Applications still need to manage their own checkpoints and avoid duplicate actions when restarting.

Understanding the Harness Evolution: Renaming in Copilot Studio, Shared Capabilities, and Expansion into Microsoft Foundry.

Why I Find This Interesting

When I first explored harnesses, the discussion was mainly about choosing the right approach for building an agent.

The Standard harness was useful for more structured, predictable workflows, while the GitHub Copilot harness supported more flexible, reasoning-intensive tasks.

Now Microsoft is extending that execution foundation beyond Copilot Studio into Foundry.

Microsoft also shared some interesting results from its internal evaluations:

• Changing harnesses made a greater difference to task pass rates than changing models in the scenarios Microsoft evaluated. Compared with the Standard harness, Microsoft reported improvements of 10.8 percentage points with an OpenAI model and 22.1 percentage points with an Anthropic model.
• After a failed tool call, agents using the Microsoft Copilot harness continued attempting the task in 84–91% of evaluated cases, compared with fewer than 7% using the Standard harness.

These are Microsoft's internal evaluation and telemetry results, not independent benchmarks. The results may vary by model, configuration and scenario, and continued attempts do not necessarily mean the overall task was completed successfully.

Still, they reinforce an important learning: a more capable AI model alone doesn't guarantee a more reliable agent. The execution system around it matters too.

A Note on Preview Availability and Cost

The Microsoft Copilot harness in Foundry Agent Service is currently in public preview.

During this preview, the harness must be selected when creating an agent. Existing agents remain unchanged, and switching the harness after creation is not yet supported.

Harness runtime usage is free until December 1, 2026, after which consumption-based runtime billing begins. However, model inference, separately billed tools and connected services may still incur charges during the free-runtime period.

This is separate from Copilot Studio's existing billing arrangements, which Microsoft says are unaffected by the rename.

Connecting This to Agent Ownership

Microsoft's Copilot Studio announcement also points to persistent, proactive agents using capabilities such as agent accounts, business process memory and triggers.

As agents take on ongoing business responsibilities, their execution capabilities must be accompanied by clear accountability.

This connects to another topic I explored recently: Who Owns an AI Agent After It Goes Live? Building the Enterprise Agent Operating Model.

An agent that can continue working, retain context and execute actions still needs a named owner, appropriate permissions, monitoring and a way to intervene when something goes wrong.

My Perspective

For me, this announcement connects the learning from August with where Microsoft's agent platforms are heading.

We started by understanding different harness types and how to build agents with them. Now we're seeing Microsoft extend a shared execution foundation across its broader AI ecosystem.

For organizations adopting agents, the starting point should remain the business scenario, required controls, complexity and cost—not simply which harness is newest.

The model provides intelligence. The harness helps turn that intelligence into controlled execution..

Related Reading — Exchange Anywhere

• Copilot Studio Harness Explained: Types, Examples, and What Each One Costs in Copilot Credits
• Copilot Studio Harnesses in Practice: Microsoft's Free Agent Academy
• Who Owns an AI Agent After It Goes Live? Building the Enterprise Agent Operating Model

References — Official Microsoft Sources

• The GitHub Copilot harness in Copilot Studio is now the Microsoft Copilot harness
• Microsoft Foundry Blog — Microsoft Copilot Harness in Foundry

Microsoft Teams in 2026: From Collaboration Platform to AI-Powered Workplace

Microsoft Teams has come a long way from chat, meetings, and calling. Looking at Microsoft's announcements from September and October 2026, it's clear that Teams is evolving into something bigger, bringing AI, intelligent communications, and stronger governance into everyday collaboration.

Here are some innovations worth watching, particularly for organizations managing Teams at enterprise scale.

Meetings are becoming organizational knowledge

One significant development is AI archives for Teams meetings. Copilot and Facilitator can use AI-generated meeting memories to answer questions about earlier discussions. These archives are stored in Teams-owned SharePoint Embedded containers without storing raw transcripts.

Deleting or expiring a transcript doesn't automatically delete its AI archive. Depending on expiration settings and Microsoft Purview retention policies, these memories may remain accessible to AI beyond the transcript's retention period. The default archive expiration is five years.

This raises an interesting question: How long should AI remember organizational conversations?

AI archives are in preview for organizations participating in the Teams Technology Adoption Program (TAP). They aren't created for certain encrypted or content-restricted meetings and don't yet support information barriers.

Other improvements include AI-generated video recaps, translated intelligent recaps, and Interpreter support for eligible Teams Rooms on Android, helping employees revisit meetings and collaborate across languages.

Copilot is moving closer to everyday work

Copilot can be opened directly from Teams messages or selected conversation text, carrying relevant context into the interaction. Teams search also helps employees discover agents and apps.

These improvements bring AI assistance closer to where employees already work.

Teams Phone is getting smarter

The Teams Phone Agent, now generally available, introduces an AI-powered receptionist that handles routine questions and routes callers with contextual summaries.

Microsoft is also expanding Intelligent Call Delegation, an AI-powered personal call assistant that can answer incoming Teams calls, gather caller context, and schedule follow-up appointments through Microsoft Bookings. Already available through the Frontier early-access program, it is planned for broader public preview in November 2026 and general availability in January 2027, subject to licensing and administrator configuration.

The Queues app is also evolving with intelligent call recaps, Copilot-assisted insights, and automatic recording and transcription.

The recording capability deserves particular attention. Recordings and transcripts are stored in SharePoint, with access governed through recording templates and the Queues app.

An important governance consideration: Explicit recording consent must be turned off for automatic call queue recording to work. Callers instead hear a recording announcement. Organizations should review applicable legal, privacy, and compliance requirements before enabling it.

Deployment requires conference mode, group- or channel-based call answering, and Shared Call History. Recording access settings and the SharePoint destination are fixed when the template is created. Changes to SharePoint structures or permissions can also cause recording failures.

Security and management are evolving too

Teams provides external bot detection and lobby controls, requiring organizer approval before detected external meeting bots can join.

The Security Detection Report in the Teams admin center provides visibility into messaging threats such as impersonation, malicious URLs, and potentially dangerous files.

Microsoft is also improving cross-tenant meeting audit visibility and app and agent installation management across Teams, Outlook, and Microsoft 365 Copilot.

My perspective

What interests me most is how Teams is becoming a place where organizational knowledge is created, remembered, and acted upon.

That creates opportunities, but also responsibilities around information protection, retention, licensing, consent, and service ownership.

These developments connect to the broader questions in The Enterprise of 2035, particularly how AI changes organizational knowledge, decision-making, and human accountability. For the practical side, Who Owns an AI Agent After It Goes Live? makes the case that ownership, access controls, and governance need to be settled before AI becomes part of everyday operations.

The real value isn't enabling every AI feature. It's making collaboration simpler while maintaining governance and accountability.

Feature availability and licensing vary. Microsoft's What's New in Teams roundup now follows a quarterly schedule, with the next edition expected in December 2026.

Official Microsoft References

• What's New in Microsoft Teams — August–September 2026
• Help customers get answers and reach the right support with Teams Phone Agent
• AI Archives for Microsoft Teams Meetings
• Manage External Bots in Teams
• Automatic Recording for Teams Call Queues
• Intelligent Recap for Teams Calls and Meetings
• Microsoft Teams Release Notes
• Microsoft — Focus on the Teams Phone calls that matter most with intelligent call delegation