It is not the tools. Most leaders are stuck at the shallow end of what AI can do, and the deep end is a skill they already have.

PwC asked more than 4,000 CEOs what their AI investment has actually produced. Fifty-six percent reported no measurable return, no higher revenue and no lower cost. PwC's own chairman summed it up: leaders have forgotten the basics.

He is right, though I would put it differently. The 56% are not behind on the technology. They are using a sliver of what it does, and most of them do not know there is more.

Start with the level almost everyone is at. You have picked up a few good prompts. Maybe you have told ChatGPT a little about yourself. So it feels like the AI knows you, and it gives you a better email, a cleaner summary, a faster draft. But a handful of prompts and a paragraph about your business is a long way from it actually understanding you, and a better draft is still a draft. The work on your plate is the same size at the end of the day.

Now picture the level above it. The work itself, finished. Before you wake up, a brief is waiting in your inbox: what actually needs you today, who you are meeting and the context on them, the two emails worth answering first, with a reply already drafted on each, for your edit and your send, never sent without you. While you slept, something read your market and flagged the one competitor move that matters this week. Before your afternoon client call, a one-page prep is ready, every note from your CRM, the open issues, what they care about.

That is not an assistant. It is closer to having a few sharp specialists working part-time through the night, the equivalent of a researcher, an analyst, and a chief of staff, so the work is finished when your day begins. Most CEOs already have an assistant. Almost none have this.

Here are a few of the jobs you can hand over:

  • A morning brief that reads your email and calendar overnight and separates the signal from the noise.
  • A research analyst that scans your competitors and your industry every day, then once a week tells you the trend, not the headlines.
  • Call and visit prep that pulls your full CRM history before every meeting, so you walk in ready instead of skimming notes in the parking lot.
  • A watch on your clients that flags who hit a problem yesterday and which relationships are cooling, before they become the call you did not see coming.

Each of those is one job, running while you do something else. Put a few together and you have a team.

So why do most leaders never get past the better email? The leap is not a smarter tool. It is a skill you already have, or already struggle with: delegation.

I have watched capable leaders try AI, get a generic result, and decide it is overhyped. When I look closer, they did to the AI exactly what they do to a new hire they do not trust. A vague instruction, no real context, no authority to act, no feedback, and then they judge it by the result. The AI did not fail them. Their delegation did. And the leaders who cannot get AI to do real work are often the same ones who cannot hand real work to their people. It is a management problem wearing a technology costume.

Which means the way in is familiar. You do not learn to code. You bring it on and direct it the way you would your best people. You give it real context, the kind you would give a senior hire on their first day: who the business is, who your customers are, what good looks like to you, where you are taking the company. You let the AI interview you and write it down, so it takes an hour, not a weekend. You connect it to where the work lives, your calendar, your email, your documents, which is the moment it stops being a chat window and gains hands. You decide what it can do on its own and what comes back to you for a yes, loosening the leash as it earns trust. And you tell it what worked, so it remembers and improves. That last part is what turns a tool into something that compounds.

 

Allie Miller, who advises some of the largest companies in the world on this, names the real failure plainly: leaders prioritize the technology over the people, even though the people side takes far more work. The ones who get real results stopped shopping for tools and started treating this as a change in how the work gets done.

None of it requires an engineer, and your size is an advantage. The hardest place to do this is a 100,000-person enterprise. A focused owner with a few hours can build what would have taken a team two years ago.

I did not believe that until I lived it. I went from using AI to write, to working alongside a system that runs real parts of my week. It changed how I operate. If I can get here, so can you.

So start small enough to begin on Monday. Take one of those jobs, the morning brief is the easiest, give the AI the context for it, connect the one tool it needs, and let it do the work and bring it back. Then tell it what to fix. That is the whole leap, in a single task. You add the next job when the first one starts saving you time.

I hold one belief, the same one behind everything I do. Every president and CEO deserves the chance to operate at a level they did not know was open to them. This is one of those levels, and it is closer than it looks.

What is the one job you would hand over first, if you trusted it to actually do the work?

Frequently asked questions

Why do most CEOs get no return from AI?

In PwC's 2026 Global CEO Survey, 56% of CEOs reported no measurable return from AI, no higher revenue and no lower cost. The reason is usually not the technology. Most are using a sliver of what it does, stuck at AI that knows them (a better email) instead of AI that works for them (the work finished). The leap between the two is delegation, a leadership skill they already have.

What is the difference between AI that knows you and AI that works for you?

AI that knows you takes some context and gives you a better email, a cleaner summary, a faster draft. The work on your plate stays the same size. AI that works for you finishes the work: a brief waiting before you wake, research done overnight, a one-page prep ready before a client call. Done well, it gives you an hour or two back every day.

Do I need to be technical or learn to code?

No. You bring the AI on and direct it the way you would a sharp new hire. You let it interview you and write down the context, you connect it to where the work lives (calendar, email, documents), and you give it feedback. A focused owner with a few hours can build what would have taken a team two years ago.

Is it safe? Will the AI act or send things without me?

You decide what it does on its own and what comes back to you for a yes. Let it read, research, and draft freely. Make it ask before it sends anything, deletes anything, or touches anything sensitive. It is progressive trust, a dial you loosen as it earns it, not a switch. The decisions that matter most still come to you.

How do I start, and how long does it take?

Start small enough to begin on Monday. Take one task you do every week, the morning brief is the easiest. Spend about an hour giving the AI the context and connecting the one tool it needs, let it do the work and bring it back, then tell it what to fix. You add the next job when the first one starts saving you time.

Source: PwC 29th Annual Global CEO Survey (2026), which found that 56% of CEOs saw neither revenue gains nor cost reductions from their AI investments; PwC global chairman Mohamed Kande noted that leaders have "forgotten the basics." Verified June 2026. The delegation flywheel builds on Allie K. Miller's self-learning organization model.

Cole Dolny is the founder of 6S Advisory Inc. and a TEC Canada Chair serving growth-minded business leaders. He spent 28 years as a partner, President, and CEO before this work. He helps CEOs and owners build healthier, more profitable companies, and understand how to put AI to work for them, not just alongside them.