Insights | Thirdera

A New Era of Visibility: Why ServiceNow’s CMDB is Essential for AI & ML

Written by Nate Aiken | Feb 3, 2025 9:17:49 PM

As artificial intelligence (AI) and machine learning (ML) continue to transform entire industries, organizations increasingly need robust governance and visibility into these powerful technologies. While traditional Configuration Management Databases (CMDBs) often tracked servers, network equipment, and application software, they did not always capture the nuanced nature of AI/ML components, such as specialized hardware or cloud-based inference services. Recognizing this gap, ServiceNow recently introduced new CMDB classes specifically designed to reflect the evolving AI and ML ecosystem. By taking advantage of these updates, your organization can establish a clearer picture of how AI-driven workloads operate - and ensure they are aligned with your broader IT service management goals.

 

More than Asset Management

Historically, CMDBs were built around physical servers and software installations. However, AI/ML workloads introduce complexities that go well beyond standard compute resources. Models rely on large datasets, specialized processing units, and sometimes ephemeral cloud-based services that make traditional asset tracking feel inadequate. This can lead to fragmented visibility, ambiguous ownership, and inefficient change management processes when dealing with AI systems.

ServiceNow’s latest CMDB enhancements address these challenges head-on. By adding specialized classes that capture the unique attributes of AI/ML applications and hardware, the CMDB now does a better job of describing modern workloads and their underlying relationships. Rather than treating AI resources as standard servers or generic software, these new classes enable organizations to track the exact nature of AI/ML processing—right down to the GPU details or cloud-based function endpoints.

 

The debut of new CMDB classes

One of the most significant developments in ServiceNow’s CMDB is the introduction of classes that speak directly to AI and ML environments. Each class focuses on a different facet of AI/ML operations, ensuring a comprehensive way to document—and manage—your intelligent services.

  1. cmdb_ci_processing_unit
    This class establishes a foundational blueprint for processing units of all kinds. Think of it as the “base layer” that can capture details relevant to CPUs, GPUs, or any specialized chipsets used in modern computing. By encapsulating core attributes (like clock speed or core count) and shared functionalities, cmdb_ci_processing_unit simplifies the process of adding new hardware types later on, ensuring your CMDB remains scalable and adaptable as technology evolves.
  2. cmdb_ci_gpu
    Building upon the base processing unit class, cmdb_ci_gpu dives into the specifics of graphics processing units—often a critical asset for AI and ML workloads. While GPUs historically handled graphics rendering, they’re now equally prized for their parallel processing capabilities. This class documents GPU-specific attributes, including the number of cores and available GPU memory. The ability to track which GPUs are installed where—or how they’re being used—helps IT teams allocate resources more effectively and anticipate future capacity needs.
  3. cmdb_ci_function_ai
    As more organizations adopt cloud-based AI services, the cmdb_ci_function_ai class serves as a structured way to track AI SaaS offerings. These often include scalable, on-demand services for machine learning, natural language processing, or data analytics provided by public cloud vendors. By distinguishing AI SaaS from on-premises solutions, your CMDB clarifies not only who owns the service contract but also where the data resides, how it’s being processed, and any regulatory implications tied to cloud usage.
  4. cmdb_ci_appl_ai_application
    Whether running in a traditional operating system environment, a container platform like Docker, or an orchestration framework such as Kubernetes, AI applications often need to be tracked at the software layer. The cmdb_ci_appl_ai_application class captures the details of these AI-driven tools—encompassing everything from the operating system it runs on, the libraries and frameworks it requires, and the models or data pipelines it relies upon. By treating AI applications as first-class citizens in the CMDB, organizations can better understand how changes to code or container configurations might impact production systems.

To obtain these class updates, visit the ServiceNow Store for the latest “CMDB CI Class Models” (v1.68.0 or newer)

 

The benefits 

The new classes not only make it easier to describe AI/ML assets in the CMDB but also provide more meaningful relationships between these assets. This is especially valuable for:

  1. Improved Visibility into Dependencies
    AI workflows often span multiple layers of infrastructure, from specialized hardware (GPUs) to cloud-based inference endpoints. Being able to visualize how an AI application (cmdb_ci_appl_ai_application) depends on a GPU (cmdb_ci_gpu) or a cloud function (cmdb_ci_function_ai) allows teams to see the bigger picture at a glance. This visibility can be the difference between a quick fix and a lengthy incident when troubleshooting performance issues.
  2. Proactive Resource Planning
    By tracking the usage patterns and configurations of GPUs or AI cloud functions, organizations can plan for future demand. If your data science team plans to scale up training tasks, the CMDB can flag insufficient GPU capacity or identify potential bottlenecks. This foresight helps prevent unplanned downtime or performance bottlenecks that could derail critical AI/ML initiatives.
  3. Enhanced Compliance and Auditing
    Data governance is a growing concern, especially when AI models ingest sensitive information. With distinct classes for AI workloads, you can quickly pull compliance-related information—such as where models are being hosted, how data is being processed, and which cloud services are in use. This granularity streamlines internal audits and ensures your organization is prepared to demonstrate regulatory adherence, whether it’s GDPR, HIPAA, or industry-specific rules.
  4. Streamlined Change Management
    Updating an AI model is not just a matter of rolling out new software. It can involve retraining on different datasets, reconfiguring GPUs for higher throughput, or switching cloud providers for better latency. When these changes are recorded in the CMDB, teams can assess potential impacts to downstream systems or dependent applications. This approach reduces the risk of production outages caused by unpredictable AI behavior or misaligned system requirements.

 

How to take advantage

To fully leverage ServiceNow’s updated CMDB, start by identifying the AI/ML resources currently in your organization. Map out who owns each piece of the puzzle—hardware, models, data pipelines, and cloud services—and then align these items with the new classes. If, for example, your data scientists rely heavily on GPU-accelerated clusters, make sure all relevant GPU details (e.g., memory, core count) are captured under cmdb_ci_gpu. If your application uses AWS or Azure-based AI services, catalog those endpoints under cmdb_ci_function_ai to maintain a record of where cloud-based inference is happening.

Automation is often key to keeping this information fresh. As AI/ML environments evolve, new containers spin up, GPUs get redeployed, and subscription tiers for cloud services change. Consider using ServiceNow Discovery or custom integrations to automatically sync updates back into the CMDB. This reduces the administrative burden on your teams and helps ensure that the CMDB remains an accurate reflection of reality.

Equally important is weaving AI/ML configuration management into your existing governance and operational processes. Anytime a production model changes, or a new GPU is purchased, relevant stakeholders—such as data science leads or finance managers—should be notified. By making these steps part of your day-to-day workflows, you’ll avoid the common pitfall of treating the CMDB as an afterthought, updated sporadically and inevitably falling out of date.

 

A source of strategic insights

ServiceNow’s new CMDB classes are not just a tactical tool for better asset tracking; they also serve as a strategic foundation for navigating the AI-powered future. As your organization’s AI footprint expands to encompass advanced analytics, natural language processing, and edge computing use cases, having a well-structured CMDB ensures that you’re not caught off-guard by complexity.

You’ll be able to answer critical questions - such as how data compliance obligations differ by AI environment or where to allocate budget for GPU upgrades - and you’ll do so with accurate, real-time information. By grounding these decisions in a CMDB that reflects AI and ML as first-class resources, you’ll foster collaboration between IT teams, data scientists, and business leaders who all rely on consistent data to guide their actions.

 

The CMDB in an AI Era

With ServiceNow’s latest CMDB updates, organizations can bridge the gap between traditional configuration management and the specialized demands of AI/ML. By accurately tracking these new classes of CIs, you create a unified environment where all stakeholders can see the dependencies, costs, and compliance implications tied to AI initiatives. More importantly, you lay the groundwork for responsible, scalable growth in the face of what is likely to be an even more AI-centric future.

If your organization wants to harness the full power of AI while maintaining clarity and control, embracing these CMDB enhancements is an essential next step. Until you're ready though, you can revolutionize your traditional CMDB management practices using AI. Our CMDB AI Advisor app on the ServiceNow Store provides real-time snapshots and intelligent historical analysis into your infrastructure. Gain instant insights into configuration items and their statuses, empowering your team to make informed decisions and maintain peak system performance.