10 Best Practices to Prepare for AI-based IT

It’s hard to read or listen to anything about IT service management (ITSM) right now without the mention of artificial intelligence (AI). Since the media success of ChatGPT in 2023, ITSM tool vendors have been adding AI capabilities to their ITSM tools at pace. Terms such as generative AI (GenAI) and Agentic AI have become commonplace in ITSM conversations. LinkedIn has seen an explosion of people willing to help your organization benefit from AI.

There’s much to consider regarding AI opportunities in ITSM and wider IT areas. Probably too much, IMHO. That’s why I’ve written this blog. I’m still a big believer in the “keep it simple, stupid” (KISS) approach to learning (and working), so my latest blog shares some of the key best practices for getting your IT organization ready for AI.

There will likely be little “rocket science” in what I write below. Instead, my blog will take a practical approach to the strategic imperative of corporate AI adoption. Importantly, stressing that integrating AI into your corporate IT environments requires so much more than selecting the right technologies and tools. Instead, preparing for AI-based IT requires:

• Detailed and thoughtful planning
• Operational readiness assessment and actions
• A shift in corporate culture to embrace the opportunities AI brings.

So, please bear this in mind when reading the following ten best practices to help you and other IT leaders prepare for AI-based IT.

1. Create a clear AI vision that’s aligned with business goals

I think we’ve all heard the horror stories of new technology being implemented just because it was available (and exciting). And if there was ever a technology that people want to experiment with, it’s AI. It’s a double-edged sword, though – there’s a lot of interest in the potential of AI to improve IT operations and business outcomes. However, it’s so, so important to avoid the “we implemented it because it was available” approach.

So, what do you want to achieve through AI? Not generics, such as greater efficiency. But, specific improvements that not only elevate IT operations and end-user experiences but also improve business results. I can’t tell you what these are for your organization. Instead, you’ll need to speak with your business peers about the better outcomes they need and then work backward to understand how your IT capabilities can help to deliver them.

This activity will help you to understand what’s needed from IT and how AI will help. For example, your business peers might want greater employee productivity from better IT services and more efficient IT support. Or they might want reduced IT costs or better employee experiences. It will likely require negotiation and compromise, but this is the only way to ensure that your plans for AI adoption align with business goals.

2. Invest in AI governance and ethics early

2025 is the year that AI governance and ethics really came to the fore. It’s not that this need has been seen as unimportant to date; it’s just climbed to the top of IT must-do lists in the last twelve months. You’ve likely noticed if you’ve attended an ITSM conference recently.

It’s because AI use raises important questions related to trust, transparency, and accountability. To address these questions and protect all stakeholders, your organization needs governance frameworks that define how AI models and capabilities are trained, validated, and monitored. This includes policies related to data privacy, algorithmic bias, and human oversight.

3. Assess your IT organization’s AI readiness

It’s a “no-brainer,” really. You can’t set off on an improvement journey without knowing where you’re starting from (and in the context of where you are heading). A quick tip from me – if you plan on leveraging ITSM tool AI capabilities in the first instance, ask your friendly ITSM tool vendor to help assess your organization’s AI readiness.

This should not only assess your current capabilities but also determine where AI capabilities can add the most value in the context of your current and potential maturity. Here, AI readiness assessments or digital maturity models can be used to baseline your current state by looking at existing workflows, transaction volumes, knowledge base maturity, existing automation capabilities, and other factors that will influence your organization’s AI success.

4. Build strong data foundations

You just knew I was going to state this! Everyone does, and there’s good reason for it. How often have you been told, “AI is only as good as the data it learns from”? By now, you might be sick of this update to the traditional “garbage in, garbage out” statement. However, your new AI capabilities must have access to clean, structured, and relevant data across various sources, such as your corporate ITSM tools (including its configuration management database (CMDB)) and monitoring tools.

Again, your friendly ITSM tool vendor will be able to help with this readiness assessment.

5. Assess and make the required people changes

As with many technology changes, AI is really a business change that affects people and processes, too. Because AI adoption will change your current ways of working. However, not only will work be done differently, which will require additional skills, but different roles (and skills) will also be needed. For example, everyone will need training to use the new AI capabilities optimally, and while GenAI capabilities will draft new knowledge articles, human effort will be required to sense-check and refine them.

6. Use AI adoption as the trigger for ITSM process redesign

There’s another old ITSM adage that “automation added to bad processes just gets to the poor outcomes faster.” In the same way, it’s important not to apply new AI capabilities to bad ITSM processes. AI and automation also offer new ways of working, rather than simply replacing the existing manual capabilities with technology.

So, when looking to add AI (and automation) to your existing ways of working, revisit your key ITSM workflows to redesign them with AI and automation integrated natively rather than as a “bolt-on.”

7. Focus on high-impact AI “quick wins” first

ITSM improvement has always called out the identification of “quick wins.” Be careful, though. Don’t get fooled into asking ChatGPT for a list of ITSM AI quick wins. Instead, focus on what matters most to your organization and its stakeholders.

Whether this is the introduction of virtual agents, recommendations for IT staff, predictive analytics, or something else, focus on the use cases that will quickly benefit your organization and provide great promotional “ammunition” for future AI investments.

8. Think carefully about how to best leverage AI in self-help strategies

You probably don’t need me to tell you of the issues IT organizations have faced in the last decade related to their IT self-service capabilities. IT self-service portals were touted as a “better, faster, cheaper” approach to IT support. But, sadly, the employee uptake and return on investment (ROI) were often way below expectations.

Now, virtual agents can leverage natural language processing and automation to empower end-users to help themselves via various channels. If done right, and please see the next point on this, AI-powered self-help will improve resolution times and employee satisfaction while significantly reducing ticket volumes.

9. It’s important to measure end-user experience, not just operational efficiency

While AI adoption might be seen as an “operational efficiency play,” if there’s one thing the ITSM industry has learned from its IT self-service portal issues, it’s that if it’s easier to call the IT service desk than use the new technology, people will. So, design your AI-powered ITSM capabilities with people in mind.

But how do you know what your end-users want and expect from their IT services and support? This is where experience management capabilities come to the fore. While traditional IT metrics usually focus on operational performance, experience-focused metrics (usually called XLAs) report how people feel about their experiences with AI, allowing your IT organization to identify and address pain points to improve its AI capabilities, adoption levels, and benefits.

10. Pilot and iteratively grow your AI capabilities

These days, no matter the change, the best practice advice is usually to pilot the proposed new capabilities, learn what works and what doesn’t (through end-user feedback), and make changes as necessary. This avoidance of what was traditionally called the “big-bang approach” prevents issues from being deployed to a wider audience and ultimately delivers better solutions. Plus, the early wins are used to improve end-user buy-in.

How are you adopting AI in your organization? Let me know in the comments.