AIOps adoption challenges for CIOs.

Apr 12, 2021. By S V Aditya

Tweet Share Share

AIOps adoption challenges for CIOs

AIOps is the new buzzword in IT operations transformation. Many articles are writing great lengths of content on the advantages of AIOps and what it can do for your enterprise. However, there is not enough content on the challenges of AIOps adoption and the realities of integration. In this blog, we shed light on the challenges of AI adoption in your enterprise.

1. AIOps skilled resources are hard to find.

To succeed in building AIOps in your organization, you need personnel with a combination of AIOps skillsets. On the one hand, you need data science and machine learning expertise. On the other, you want someone with years of experience in IT Operations who understands its challenges and has implemented IT best practices in their organization. These skillsets are hard to find, making the native development of AIOps solutions a long, slow process in most companies.

CIOs facing this challenge need to consider if their organizational goals lie in building a solution or buying/partnering with a provider. While building AIOps solutions might be costlier, it can allow for greater customization and control. Buying or Partnering is a cheaper alternative with the added benefit of long-term support for platforms.

2. AIOps integration with legacy IT infrastructure is difficult

Many organizations have grown inorganically by acquiring other companies. Consequently, they also integrated their legacy IT infrastructure. Usually, it is cheaper to work with legacy systems than to overhaul the entire infrastructure. As organizations increasingly took this approach, they created data silos with these legacy systems that do not support easy integration with other monitoring tools. This often leads to relying on other tools for deeper information on these systems, cutting down on AIOps platforms' value.

AIOps solutions must have built-in mechanisms to add collectors without many manual workarounds to get around this challenge. Solutions that use internal data lakes for KPIs, metrics, and logs instead of relying on external systems are also much more effective.

3. AIOps solution setup and configurations are lengthy

Let's face it - you need a solution that can monitor the entire infrastructure stack. To do that, you need to be able to set up configurations for monitoring CIs across a wide variety of platforms, tools, databases, and nodes. Building such configurations and setting items like KPI-alert thresholds will be a lengthy process, even if it is a one-time issue.

You need a solution that can intelligently capture the default state of the system and build its configurations. This way it can activate features like incident recognition immediately. Such a solution can rapidly accelerate the speed of AIOps deployment from months to a week. Additionally, your AIOps solution needs to have customized templates in the form of collection groups that can be assigned to multiple CIs of your choosing at any time.

4. AIOps is more than Application Performance Management

Most vendors that present themselves as AIOps solutions are simply Application Performance Management vendors. The complexity and the value of AIOps are missing in such services. For example, an APM vendor can identify what applications are misbehaving. But an AIOps solution can identify which component is the root cause of issues and use reinforcement learning models to auto-remediate incidents with clear audit trails of actions taken and state changes in the systems.

CIOs have to consider their organization's priorities and needs before investing in AIOps tools - if they require only monitoring and observation elements or if they also need automation for cost savings. An end-to-end solution will cut down on the tool proliferation, but a single module can be a lightweight solution that integrates with your other platforms.

5. AITSM (AI for IT Support Management) is treated as a separate entity from AIOps

AIOps involves at least three elements - Observe (Monitoring), Engage (Support Management), and Act (Automation). By treating the Engage element as a separate service, most solution providers are missing out on value creation like auto-fulfillment of service requests and adding to the noise from an increasing number of tools handled by IT operations.

There is a huge potential for cost savings with L1 support automation using AIOps. Moreover, Engage covers analytics of incident reports, tickets, and issues handled by AI solutions. This information is key to having a complete picture of the value of AIOps.

6. AIOps monitoring is confined to KPIs and metrics

Most AIOps vendors are limiting their scope to metrics and KPIs, i.e, structured data. However, a lot of high-value information is stored in unstructured data like logs, traces, and ticket messages.

To be effective in predictions and incident recognition, AIOps platforms must have the capability to work with unstructured data. CIOs have to ensure that the platforms they are considering can handle such nebulous information and extract value from it.

7. AIOps tools need to demonstrate RoI and Business Value

AIOps platforms and services largely monitor technological aspects like infrastructure(CPU, Load, Memory, etc.) and application performance metrics. While this is very useful from an incident remediation point-of-view, the top brass of enterprise needs to understand the business impact to decide on AIOps investments.

CIOs must look for solutions that can map performance metrics with business value in a single pane of glass view to enable deeper insight.

8. AIOps culture has to be cultivated in the enterprise

Finally, there is the human element. Cultural inertia and skepticism to AIOps tools are primary barriers to implementation in a large-scale organization. There are factors like fear of job loss at play that make employees resistant to automation technology. Balancing employee expectations is critical when implementing such new solutions.

Executives must consider proof of concept type approaches with use cases that have clear, demonstrable value to start with. For example, a goal should be measurable, like "I need an MTTR reduction of at least 25%" after implementing the root cause analysis module. Setting clear goals shows the value of solutions that can build momentum as more employees become familiar with the technology.

Adoption of any new technology is challenging. But the most successful companies can overcome such barriers and use AIOps technology as a growth engine.

For more information on how to transform your enterprise with AIOps, please visit

Share this blog.

Tweet Share Share