An Executive's Guide to Getting Started with AI

It's not a tool. It's an employee who needs onboarding.

McKinsey just announced they now count 25,000 AI agents as part of their 65,000-person workforce. Nearly 40% of their people aren't people at all. CEO Bob Sternfels wants every human employee working alongside an AI agent within 18 months.

Read that again.

One of the most elite professional services firms on the planet isn't asking if AI will change how they work. They're counting AI as headcount.

If you haven't started using AI yet, this article is for you. And if "started" means you've typed a few questions into ChatGPT and weren't impressed, this article is definitely for you.

The Mental Model That Changes Everything

Most people approach AI like they approached Google or Excel. A tool you learn to use. Commands to memorize. Features to understand.

That's the wrong frame.

Working with AI is more like working with a new employee. A brilliant one. Someone with a photographic memory, zero ego, and the ability to work 24/7 without getting tired or resentful.

But also someone who just started. They don't know your company. They don't know your preferences. They don't know what "good" looks like in your world.

Would you hand a new hire a task on day one with no context and expect perfection? Of course not. You'd onboard them. You'd explain how things work around here. You'd give them examples. You'd correct their first attempts and explain why.

That's exactly what AI needs from you.

Why "I Tried It and It Wasn't That Impressive" Doesn't Mean What You Think

I hear this constantly from executives in my network. Smart people. High performers. They gave AI a shot, got mediocre results, and moved on.

Here's what actually happened: they hired a brilliant employee, gave them zero context, and fired them after one task.

"Write me a marketing email" with no information about your brand voice, your audience, your product, or what's worked before is like telling a new marketing hire "figure it out" on day one. The results will be generic. Useless. Disappointing.

The AI didn't fail. The onboarding failed.

The Onboarding Framework

If you've managed people, you already know how to work with AI. You just need to apply the same thinking.

This isn't complicated. But it does require you to treat AI like what it is: a capable entity that needs context to perform.

The Shift I've Noticed in My Own Work

As I use AI more, something strange has happened. I spend far more time planning than executing.

Old pattern: I'd jump into a task, figure it out as I went, iterate through trial and error. Writing a proposal? Open the doc, start typing, see where it goes.

New pattern: I spend 80% of my time getting the plan exactly right. Specifying exactly what I want. Gathering the context AI needs. Then AI executes in a fraction of the time it would take me.

The work has inverted. Thinking is the job now. Execution is delegated.

When the plan is solid, AI can produce work that's not just good. It's exactly what I would have produced if I were 100x faster and never got tired. Because it has my thinking embedded in the instructions.

The Real Risk

Here's what concerns me about executives who haven't started yet.

The risk isn't that AI will take your job. Not directly. The risk is that your competitors are using AI to serve your clients better, faster, and cheaper than you can.

While you're still doing everything manually, someone else is getting 10x more proposals out the door. Responding to emails in minutes instead of hours. Producing analysis that used to take weeks.

This isn't theoretical. McKinsey isn't counting AI agents as employees because it's a fun PR move. They're doing it because those agents are generating real output.

The gap between AI-augmented professionals and manual-only professionals is widening every month. And unlike most technology shifts, this one doesn't require capital or technical expertise to bridge. The tools are available to everyone. Claude, ChatGPT, Gemini. They cost less than a decent lunch.

The only barrier is the mental shift from "tool I need to learn" to "employee I need to onboard."

Where to Start Today

If you haven't used AI, here's your homework. Not a grand strategy. Not a company-wide initiative. Just one thing.

Take your most recent client email. The kind you write often. Paste it into Claude or ChatGPT with this prompt:

"I wrote this email to a client. I want you to learn my writing style from it. What do you notice about my tone, sentence structure, and word choice?"

Then give it a new email to write. See what happens.

If the output isn't great, don't give up. That's your training opportunity. Tell it what's wrong. Show it a better example. Iterate. Just like you would with a human employee who's learning.

The people who thrive in 2026 won't be the ones who "mastered AI." They'll be the ones who started.

I wrote more about the levels of AI adoption in my piece on why AI resistance is futile. If you want to understand where you currently sit and where you could be, start there.

But don't just read about it. Open a chat window. Start the onboarding process. Your new employee is waiting.