Transforming Microsoft’s enterprise IT infrastructure with AI

AI is changing everything at Microsoft, including our approach to core IT.

We in Microsoft Digital (MSD), the company’s IT organization, are using the advent of generative AI to reexamine and transform our entire IT infrastructure.

“We’ve crossed an important threshold with AI,” says Mark Sherwood, vice president of Infrastructure and Engineering Services in MSD. “We’re now using it to transform all our core IT services, to make everything we do more efficient, and secure.”

Sherwood and his team manage our core IT services, a massive enterprise IT estate that supports all of Microsoft’s business worldwide. Microsoft is an expansive universe of connected devices made up of hundreds of thousands of PCs and laptops, conference rooms, building IoT sensors, and personal devices—all dependent on a foundation of network connectivity and security to enable seamless access to the tools and services our employees rely on every day.

We’ve crossed an important threshold with AI. We’re now using it to transform all our core IT services, to make everything we do more efficient, and secure.

— Mark Sherwood, vice president, Infrastructure and Engineering Services, Microsoft Digital

It’s clear that AI brings immense value to our IT infrastructure.

“This is a fascinating time to be working in IT,” Sherwood says. “We’re using AI across all of our services, and now we get to take that investment to the next level. Now it’s all about seeing what we can do with it.”

Aligning IT infrastructure innovation with the rest of the organization

The strategy for AI transformation in core IT infrastructure is one part of a larger vision for the impact of AI across all of Microsoft Digital.

“The potential for transformation through AI is nearly limitless,” says Natalie D’Hers, corporate vice president of Microsoft Digital. “We’re evaluating every service in our portfolio to consider how AI can improve outcomes, lower costs, and create a sustained competitive advantage for Microsoft and for our customers.”

We’re hyper-focused on our employee experience, and AI will be instrumental in shaping the future of how Microsoft employees interact with customers, the organization, and each other.

The potential for transformation through AI is nearly limitless. We’re evaluating every service in our portfolio to consider how AI can improve outcomes, lower costs, and create a sustained competitive advantage for Microsoft and for our customers.

— Nathalie D’Hers, corporate vice president, Microsoft Digital

Transforming and securing our network and infrastructure

AI holds enormous potential across all of Microsoft Digital , but within IT infrastructure, the benefits of AI-enabled transformation play out across several specific pillars where we’re focusing our efforts: Device management, network infrastructure, tenant management, security, and the IT support experience.

Security

We can’t transform without adequate security. Properly implemented security controls and governance provide the secure foundation on which our engineering teams build solutions, and that security is especially relevant as we incorporate AI into our services and solutions.

Securing our network and endpoints is imperative, and our Zero Trust Networking efforts across our IT infrastructure provide essential protection against threats to our network security. AI will enhance the security and compliance of these efforts in our cloud and on-premises environments.

AI-based network assignment for devices will simplify network classification and provide more robust risk-based isolation to isolate risky devices and reduce unwanted movement across the network.

We’re automating access controls for our wired and wireless networks to improve security effectiveness. AI-infused processes for analyzing device vulnerabilities, detecting anomalous firewall traffic flow, and diagnosing other network incidents will play a critical role in our continued shift toward the internet as our primary network transport.

We anticipate that AI-supplemented capability in Microsoft 365’s multi-tenant organization feature will help us meet our ever-changing network segmentation needs by maintaining tenant separation and enabling secure tenant cross-collaboration when required.

AI will help us manage third-party app access and revolutionize how we understand user interactions with applications across managed devices or SaaS platforms. We’ll increase access efficiency and reduce costs by capturing third-party app usage and needs more accurately, using AI to determine the how, why, and when of user access.

Intelligent infrastructure

Sherwood (left to right), Apple, Selvaraj, and Suver appear in a composite image. — Mark Sherwood (left to right), Pete Apple, Senthil Selvaraj, and Phil Suver were part of the team incorporating AI into Microsoft Digital’s vision for core IT.

Software-defined networking and infrastructure code are already transforming how we approach networking, but AI amplifies the benefits radically.

AI enables us to build data-driven intelligence into network infrastructure, engineering, and operations. AI-driven processes will help us eliminate configuration drift, comply with security policies, reduce operator errors, and efficiently respond to rapidly changing business needs.

We’re implementing AI-driven automation to simplify resource management and deployment, capitalizing on the flexibility provided by software-defined networking and infrastructure as code.

AI will assist with generating code designs, defining and managing network configurations, managing deployments, conducting pre- and post-deployment verifications, and assisting with change management over time. Near real-time streaming telemetry from network devices will form the foundation to guide operation and continuous improvement.

We’re improving network self-healing capabilities by using AI to detect and remediate network issues, creating a more reliable, resilient, and elastic network environment and reducing human intervention and potential for error.

One of our current projects is creating an AI-based assistant app for our direct engineering teams that mines and analyzes our current network infrastructure catalog, providing an advanced set of capabilities that supplement our engineers’ expertise in the field. The assistant app improves productivity and mitigation time for network infrastructure incidents. The AI component is trained on more than 200,000 prior incidents for anomaly detection and predictive analytics. We’re confident it will lead to a considerable reduction in network outages and maintenance costs.

Device management

With more than 1 million interconnected devices, AI-powered capabilities will significantly benefit our device management practices with a focus on user and administrator workflows.

We’re implementing intelligent device recommendations to ensure our employees have the best tools to do their work. Building AI into a centralized device lifecycle management tool will create efficiencies in procurement, tracking, and responsible device recycling.

We’re designing AI-powered predictive maintenance and intelligent troubleshooting to reduce device-related issues significantly. AI-enabled device maintenance schedules and tasks will automate the device management process and reduce the load on our IT help desk by correcting device issues before they become user problems, reducing device-related helpdesk incidents.

Across our vast scope of device management, many alerts and tickets contain information or fixes that our helpdesk engineers can use in other situations. We’re employing AI to generate device insights by analyzing a massive set of signals, including device configurations, network traffic, vulnerabilities, and user behavior. These insights will power more informed decisions across the device management portfolio, including device replacement, software updates, and capacity increases.

We have more than 100,000 IoT devices on our corporate network. AI-automated IoT device registration will create more robust and efficient IoT device management, tracking, and security.

AI and machine learning will help us to perform aggregated meetings and call data for device monitoring across personal devices, Microsoft Teams meeting rooms, networks, IoT devices, and Microsoft 365, improving and safeguarding the user experience.

Tenant management

Our cloud tenants in Microsoft Azure, Microsoft 365, Dynamics 365, and the Power Platform are among those platforms’ largest and most complex implementations. Our internal implementation includes more than 205,000 Microsoft Teams, 534,000 SharePoint sites, 430,00 Microsoft Exchange mailboxes, 93,00 Power Apps, 5,000 Viva Engage communities, and a massive 25,000 Microsoft Azure subscriptions.

It’s a lot to manage, and AI will improve how we do it.

In tenants of our size, unmanaged assets can lead to unnecessary costs. Our asset compliance and lifecycle management processes will include an AI-powered compliance assistant that informs tenant users and owners, recommends assets for deletion, and proactively identifies areas of high risk for the tenant. Through the assistant, tenant admins gain an all-up view of compliance status and can investigate and resolve issues more granularly.

AI is also simplifying and streamlining our license management processes. We adhere to precise rules and regulations, which result in complex access scenarios across different countries and regions. AI will bolster our ability to detect and remediate non-compliant tenants amidst this complexity.

IT support

We’re poised to transform how Microsoft employees interact with our support services using generative AI.

Our employees interact with Microsoft support services in a complex, global hybrid environment. Our self-help solution using Microsoft Azure OpenAI will enable contextual and human-like conversation and support in the employee’s local language. Our chat and incident summarization tools will use AI to summarize incidents and provide context when assisted support is necessary.

We’re infusing our support ticketing systems with AI capability for forecasting support requirements and proactively checking the health of devices to reduce issues and improve resource planning and response times.

Transforming our IT infrastructure as Customer Zero

As Customer Zero for Microsoft, we pilot and deploy new products and capabilities in our IT infrastructure before releasing them externally. Our scale, size, and knowledge of our products and services enable us to envision connected experiences across large enterprises, manage complex combinations of product use cases, and engineer solutions on top of our product platforms.

AI improves our role as Customer Zero by accelerating insights and improving time-to-value. We’re using AI capabilities to capture, review, analyze, and report on the most important and actionable insights from the Customer Zero experience. We’re also using AI to redevelop processes, regulatory compliance, security reviews, and deployment practices within the Customer Zero environment.

We will continue to infuse AI into every dimension of our enterprise portfolio. We’ll continue to identify new opportunities for building AI-powered applications and services that improve how we deliver IT services to the company.

— Mark Sherwood, vice president, Infrastructure and Engineering Services, Microsoft Digital

Looking forward

It’s almost impossible to envision a future for corporate IT infrastructure without AI. Our active planning for AI in our infrastructure is continually evolving, and we’ve only just begun our implementation. We’re positioning Microsoft to be a catalyst for innovation, and we’re committed to innovating with AI to streamline our IT operations.

“We will continue to infuse AI into every dimension of our enterprise portfolio,” Sherwood says. “We’ll continue to identify new opportunities for building AI-powered applications and services that improve how we deliver IT services to the company.”

By showcasing our progress with AI capabilities, we aim to transform our approach to AI internally here at Microsoft and to fuel a similar transformation across the IT sector.

Here are four important steps you can take to transform your IT infrastructure with AI:

Make device handling smarter with AI. Use AI to manage all devices better, helping to fix problems before they affect people and easing the workload for your IT team.
Use AI to improve the network. Integrate AI into the network system to make it more intelligent and more adaptable, which helps reduce downtime and facilitates faster and easier changes.
Manage cloud services better with AI. AI can help keep track of cloud services, ensuring everything is used properly and securely.
Boost security and helpdesk with AI. Enhance safety and helpdesk services using AI, leading to better network protection and quicker, more effective support for employees when they need it.