You run a sports organization, lead a department or manage relationships that directly affect the business.

You probably don’t have time to keep up with every AI model, new feature or prompting technique. And I don’t think that’s where your time should go.

But there’s a difference between knowing how to use ChatGPT and knowing how AI could improve the way you work. The second one matters more. Four questions cover most of it.

1. Where could AI actually make my job easier?

One-off tasks can benefit from AI, but recurring responsibilities are often where the biggest opportunities are. Preparing for leadership meetings, reviewing department updates, organizing briefings, researching potential partners and preparing follow-ups.

Using AI for a task is asking ChatGPT to summarize a document you’ve supplied. Improving a workflow is a repeatable process that prepares it the same way every time, with the right information, instructions and review. Neither needs sophisticated automation.

Consider preparing for a weekly leadership meeting. A hockey executive needs updates, outstanding decisions, background and an agenda. Assembled manually, that’s searching emails, reports and old notes. A repeatable workflow could organize information you provide through authorized channels, flag outstanding items and prepare the briefing, follow-up drafts and action items for your review. Priorities, decisions and people stay yours.

The opportunity isn’t more things to ask ChatGPT. It’s the repeating work, and whether there’s a better way to get it done.

2. Why doesn’t AI understand how I work?

It doesn’t automatically have the context it needs about your responsibilities, priorities, communication style or what a good result looks like. Depending on the tool and how it’s configured, some information can be retained or made available across interactions, but that doesn’t mean it has a complete understanding of your organization.

Workflows need consistent instructions, background, approved reference materials, examples of outputs, access and a review process.

If you’re explaining your role, your organization and what a good result looks like on every use, you’re doing work you shouldn’t have to repeat. Some of that context can be built into a repeatable process. That doesn’t mean AI suddenly understands your business, just that it starts from a better place: context improves usefulness without guaranteeing accuracy, and confidential information shouldn’t go into these tools without the appropriate permissions and safeguards.

3. What should still be left to me?

AI can prepare information, identify questions, compare materials you’ve supplied and draft communications. Deciding is different.

Consequential decisions belong to you: commercial negotiations, sensitive personnel matters, confidential player information, organizational reputation, important relationships, contractual commitments. Not because AI couldn’t produce an answer, but because accountability for those calls can’t be delegated to a tool.

AI can help you prepare for a difficult conversation. It shouldn’t decide how that conversation is handled. It can organize information before a negotiation. It shouldn’t set the value of the relationship or commit your organization to terms.

The opportunity is to reduce the work around your decisions, not outsource them. Lower-risk steps don’t need your approval each time. Oversight should match the stakes.

4. Is this actually saving me time?

Using AI often isn’t the same as working more efficiently, and a generated document isn’t automatically an improvement.

What matters is the whole process, preparation, review and corrections included. If AI saves 20 minutes of drafting but adds 15 minutes of editing, that’s a fairly small improvement. If a process that used to take an hour can reliably be completed in 20 minutes, that’s a different conversation. Those numbers are illustrative. The principle isn’t.

The point isn’t how often you’re using AI. It’s whether the work is improving, in time, consistency or quality. I’ve written elsewhere about measuring that properly. Here it’s simpler: if the work isn’t getting easier, the setup needs another look.

So what do you need to understand? Enough to identify opportunities, ask questions about information, security and access, know where human oversight is necessary and judge whether the outcome justifies the effort.

You don’t need to master advanced prompting, configure every tool, build integrations, troubleshoot technical problems or maintain every workflow. Some executives enjoy building their own. It isn’t a prerequisite for benefiting from AI.

Knowing where AI belongs in your work is a leadership responsibility. Personally configuring every solution doesn’t have to be.