How to upskill in AI: Lessons from a CIO

CIOs don't run every AI upskilling program, but they define the strategy and guide IT.

At Sedgwick, a global claims management company, CIO Sean Safieh uses Sedgwick University, a customized learning platform, to help employees learn new AI skills and apply them in their work. He encourages employees to become comfortable with AI copilots and code development tools, creates safe spaces for experimentation and offers multiple learning formats, including instructor-led and self-guided sessions. For CIOs navigating AI upskilling, his approach shows how to lead from the top and give teams the space to learn and apply AI effectively.

In the following interview, Safieh explains how Sedgwick balances structured training with hands-on learning and how to set expectations across IT and the broader workforce.

Editor's note: The following transcript was edited for length and clarity.

Where did you start with AI upskilling -- within IT or the broader workforce?

Sean Safieh: A little bit of both. We began our AI journey years ago, before generative AI [GenAI] became the big focus. Early on, machine learning was one of the first areas we explored, led by our data science team.

As GenAI tools emerged, we began introducing them more broadly across the organization. We used Sedgwick University, our webinar-based training platform, along with vendor-led training to help employees understand how these tools could support their workday. It started with simple examples, like using AI to write an email or review your calendar and identify the most important meeting. The goal was to help people learn how prompting works.

On the IT side, we focused on how to use AI code development tools to support development and QA. In that context, generative and agentic AI help developers with everyday tasks. A lot of the focus is on administrative work that developers don't enjoy doing -- things like documenting technology or writing release notes. We also encourage employees to share ideas and challenges through internal channels and portals, which helps drive broader adoption across the organization.

Is Sedgwick University just for IT or the broader workforce?

Safieh: Both. It's our talent portal, which lets employees choose courses and learning paths across a variety of topics. So, as an IT developer, you can learn specific coding languages and tools at your own pace. There are several topics across technology, operations and the systems that our teams use.

Does Sedwick University sit atop a third-party learning platform?

Safieh: Yeah. Our HR team came up with the idea of creating a Sedgwick University, and within that university, we partner with a third-party tool that offers some of its own course-based training. But we can also add our own training videos and learning materials to that site. So, if we're doing something proprietary that we want our teams to learn about, we can create it internally and upload it so all our colleagues and team members have access to it.

How else are you training people in AI?

You'll have some people who love to watch computer-based training, some who love to sit in a room with an instructor and some who like to read.

Safieh: We also have vendor-led training and sometimes third parties come in on specific topics. If we're deploying a new AI tool and want more advanced learning around it, we can bring folks in to help with that. It's not a one-size-fits-all approach.

Whether in IT or any other department, you'll have some people who love to watch computer-based training, some who love to sit in a room with an instructor and some who like to read. Then there are those like me who would rather play with it, learn on their own and read documentation if they need it. We have all those capabilities built into Sedgwick University, along with the other materials and vendors we can use.

For most companies that aren't hyperscalers like Google or Microsoft, how deep does AI expertise need to go within IT?

Safieh: It depends. There are two tracks I can talk about. One is the folks developing and integrating the GenAI tools that our teams use. Typically, they need a deeper level of expertise. We often use third-party tools, but those teams still need to understand prompting and how the tools work so they can integrate them into the end-to-end experience for the user.

Then, you have people using those tools in their day-to-day work -- for example, developers using code development tools. In those cases, they need to understand how to incorporate the capabilities into their workflow, whether that's simple prompting or using agentic AI as part of their process. It also comes down to trust -- learning when to trust the output and how to validate it.

Many of these tools are now built directly into software people already use. Generative and agentic AI tools have made it much easier to get started. A lot of our focus has been on giving employees quick tips so they understand what the tools can do, while also giving them the space to learn.

What has worked better in practice: structured training or learning through live projects?

Safieh: There's no one-size-fits-all. Structured training is great for giving people an entry point -- understanding how the technology works, what they need to do with it and some general use cases.

But AI has become so powerful that it's difficult to train people on every scenario or use case, especially in IT, where we're using it for code development and task automation. So, structured training helps people build that baseline and get comfortable.

From there, hands-on learning becomes important. I'd also add a third element: creating channels or safe spaces where employees can talk with others who are using the tools. That allows people to share what they've learned, what isn't working and ask questions like, 'How do I do this?'

What does AI literacy look like for non-technical employees at your company?

Safieh: It starts with understanding what AI is and how it can support day-to-day tasks. It's not something that does everything for you or replaces you -- it's a tool to help elevate your work, especially around administrative tasks.

The next important piece is knowing how to interact with AI effectively -- how to ask questions, use prompts and get the most accurate responses. GenAI is very powerful, but its value depends on how you communicate with it.

How much responsibility do you, as CIO, have in training the larger workforce?

Safieh: We have a training and talent team that drives Sedgwick University. That team handles most of the formal training, but there's also a need within each department -- including IT -- to educate our own teams on what AI is, why we're using it, its potential and where it's headed.

As leaders, it's important to lead from the top. If we aren't talking to our employees about AI, it can feel taboo. If we believe these tools are vital to our organization's growth, then we must hold our teams accountable for learning them. Technology changes fast, and understanding why and how we're adapting as an organization is just as important as knowing the tools themselves.

What matters most is the ability to learn, be agile and fail fast, no matter your role.

Are any IT roles becoming more important because of AI?

Safieh: No specific role is more or less important right now. What matters most is the ability to learn, be agile and fail fast, no matter your role -- whether you're a developer, project manager or in another IT position.

Technology is moving faster than ever, and that pace will only increase. It's less about always staying ahead and more about adapting and learning as you go. Roles are shifting, and the divide between IT and business continues to blur. Deep technical skills will always be in demand, but technical skills combined with the ability to learn are even more valuable.

If a CIO called you today asking where to start with AI upskilling, what would you tell them?

Safieh: I've had this conversation a few times. First, I ask where they are. Are they enabling any AI tools today? If not, they're starting from scratch. My first recommendation is to look at copilot-type tools that let teams use large language models and GenAI at their desktops. Next, enable development teams to use code development tools. These tools are becoming very powerful, and the value teams will see is tremendous.

I also tell CIOs not to chase every new tool that comes along. The marketplace moves too fast for that. Instead, pick a few tools that are fit for purpose and make sense for your organization. Train your teams to use them effectively, build that knowledge and skill set, and then start exploring what other use cases can be addressed with additional tools or the ones you already have in place.

Tim Murphy is site editor for Informa TechTarget's IT Strategy group.