Fifty-six percent of CEOs say they’re getting “nothing” from their AI investments. That’s according to PwC’s 2026 Global CEO Survey, one of the largest annual executive surveys in the world.
Meanwhile, companies are on track to spend over $500 billion on AI this year. The tools are better than ever. The talent market is maturing. The use cases are proven.
So where’s the disconnect?
After working with dozens of organizations on AI adoption, I keep coming back to the same answer. The biggest bottleneck to your AI strategy isn’t your budget, your tech stack, or your team’s skill level. It’s sitting in the corner office.
The Executive Sideline Trap
Here’s a pattern we see constantly. An executive team decides AI is a priority. They approve a budget. They assign someone to lead it — maybe a Chief AI Officer, maybe a VP of Operations, maybe a working group. They bring in consultants. They evaluate platforms.
And then they wait.
They wait for the ROI case. They wait for the pilot results. They wait for someone else to prove it works before they touch it themselves.
This is the executive sideline trap: sponsoring AI initiatives without ever using AI yourself. Treating it like you treated SaaS adoption a decade ago… compare features, pick a vendor, roll it out, measure results.
AI doesn’t work that way. You can’t evaluate AI from a slide deck. The gap between “I’ve read about what AI can do” and “I’ve used AI to do my actual work” is enormous, and it’s a gap that no vendor demo or analyst report can bridge.
The “show me the ROI first” stance sounds reasonable. It’s what good executives do — demand evidence before committing resources. But with AI, it’s backwards. You can’t accurately assess ROI on a capability you don’t understand experientially. You’ll either overestimate it (chasing hype) or underestimate it (dismissing real value because the demo didn’t impress you).
The data supports this. McKinsey’s State of AI research found that AI high performers are three times more likely to have senior leaders who actively demonstrate ownership and role-model the use of AI. Three times. Not a marginal edge.
And the consequences of sideline leadership are measurable. HBR reported on two banks rolling out the same AI program. At one, the VP of Technology acted as sponsor, but the CEO never mentioned the program publicly. Adoption stalled at 31%. At the other, the COO sponsored the initiative and the CEO referenced the AI rollout in multiple all-hands meetings. Adoption hit 79% with measurable ROI.
Same technology. Same type of organization. Wildly different outcomes. The variable was leadership visibility.
The Permission Cascade
When a CEO uses AI publicly — mentions it in meetings, shares what they’ve learned, talks about a workflow they’ve built — it creates what I call a permission cascade. It signals to every layer of the organization that experimenting with AI is not just allowed, it’s expected.
Research from Perceptyx found that organizations where AI adoption is driven by leadership with clear strategies report 62% of employees fully engaged with AI tools. More importantly, employees in those organizations are 7.9 times more likely to say AI has positively impacted their workplace culture compared to organizations without a structured, leadership-driven approach.
The reverse is equally powerful. When leaders don’t use AI, middle managers read the signal correctly: this isn’t really a priority. Pilots get quietly deprioritized. The “AI initiative” becomes another line item that nobody wants to kill but nobody champions either. Six months later, the CEO is in the 56% wondering why they got nothing from their investment.
We saw this play out with a client last year. The CEO was skeptical but willing — a common starting point. After we helped him build a personal AI workflow for his weekly board prep, he started dropping references to it in leadership meetings. Nothing forced, just natural mentions: “I had Claude summarize the customer feedback trends and something interesting came up.” Within a month, three of his direct reports asked for help building their own workflows. Within a quarter, the company had moved from scattered experimentation to a coherent AI adoption plan.
The Aha Moment — And How to Engineer Yours
AI has a before/after moment. We call it the aha moment, and most executives haven’t had it yet.
Before the aha moment, AI is an interesting technology that other people use. After it, AI is something you can’t imagine working without. The shift is the moment you realize this tool just did in 90 seconds what used to take you 45 minutes, and it did it well.
For most executives, the aha moment comes from a specific kind of use case: synthesizing information you’re already drowning in, like the firehose of communication that hits you every day — Slack threads, email chains, meeting notes, weekly reports from six different teams.
Imagine starting your morning with a two-minute briefing that pulls together everything important from the last 24 hours across all those channels, organized by priority, with the three things that actually need your attention highlighted at the top. This is something you can build in 30 minutes. In fact, this is how my morning starts.
Here’s how:
Pick one AI tool — Claude, ChatGPT, whatever you’re comfortable with. I prefer Claude, especially Claude Cowork as it has agentic abilities. Connect it to your email inbox, calendar, and Google Drive. Then ask for a synthesized briefing organized by: decisions needed, risks flagged, and progress updates.
The first time you do this, two things will happen. First, you’ll see connections between threads you hadn’t noticed because they were buried in different channels. Second, you’ll realize how much time you spend every day doing this synthesis manually, poorly, in your own head while context-switching between tabs.
That’s the aha moment. And it matters strategically because once you’ve had it, you can see where AI fits across your organization. You stop asking “what’s the ROI?” in the abstract and start asking “where else are my people drowning in synthesis work?” That’s a much more productive question.
Five Things Every Executive Should Do This Quarter
Enough theory. Here’s what to actually do.
1. Use AI for one real task daily for two weeks.
Pick an actual part of your workflow like prepping for a board meeting, drafting a response to a strategic question, analyzing a competitive landscape. Two weeks is enough to get past the awkward early phase where everything feels slower. By week two, you’ll have found at least one task where AI is genuinely faster and better.
2. Build one personal workflow.
Pick something you do every week and build an AI-assisted version of it. Meeting prep, email triage, report synthesis, competitive monitoring. The key word is “personal.” This isn’t about rolling out an enterprise tool. It’s about you, with one AI assistant, making one recurring task meaningfully better. Once you’ve done it for yourself, you’ll understand what it takes to do it for your team.
3. Ask your team what they’re already using AI for.
You will be surprised. Shadow AI is everywhere in your organization right now. People are using ChatGPT to draft emails, Gemini to summarize documents, and Claude to debug spreadsheets. They’re just not telling you because they’re not sure if it’s allowed or encouraged. Bringing this into the open does two things: it gives you a map of where AI is already adding value, and it signals that experimentation is welcome.
4. Kill one process that AI makes obsolete.
This is the one most executives skip, and it’s the most important. Don’t just layer AI on top of existing workflows. Find one process — a weekly report that nobody reads, a manual data compilation, a recurring meeting that exists only to share information — and eliminate it. Replace it with an AI-powered alternative or just remove it entirely. This sends a stronger signal than any all-hands speech about innovation.
5. Set AI adoption goals that reward outcomes, not usage.
“Everyone must use AI” is a bad goal. It incentivizes performative adoption — logging in to check a box. “Reduce report creation time by 50%” is a good goal. “Decrease time-to-first-response for customer issues by 30%” is a good goal. Measure what AI enables your people to accomplish, not whether they opened the app.
The Anti-Patterns
While we’re being specific about what to do, let’s be equally specific about what not to do. These are patterns we see repeatedly, and they all share a common root: leadership that delegates AI adoption instead of owning it.
The AI Committee. A cross-functional team is formed to “evaluate AI opportunities.” They meet biweekly. They build a spreadsheet of tools. They develop evaluation criteria. Six months later, they’ve produced a thorough report and made zero decisions. Meanwhile, competitors have shipped three AI-powered features. Committees are where AI momentum goes to die. Pick a tool, set a timeline, run an experiment. You can always switch later.
The Mandate Without Modeling. “Starting next quarter, everyone is expected to incorporate AI into their workflows.” Sent via email by a CEO who has never used AI for anything. Your team will see through this instantly. Mandates without modeling breed cynicism. If you’re not using it yourself, don’t tell others to.
The Pilot Graveyard. Start ten pilots across ten departments. Give each one a small budget and vague success criteria. Check in after six months. Find that most fizzled out because nobody owned them. Declare that “AI doesn’t really work for our industry.” This is how most AI launches fail, because leadership treated AI as a standalone project rather than an operating model shift.
BCG’s research captures this perfectly: 88% of companies now report regular AI use, but only a small minority have scaled it in a way that measurably changes how work gets done. The gap between “using AI” and “getting value from AI” is a leadership gap.
Start With Yourself
You don’t need a vendor evaluation. You don’t need a committee. You don’t need a six-month roadmap.
You need 30 minutes and a willingness to get your hands dirty.
Open Claude or ChatGPT. Take your most overwhelming information stream — the one that eats an hour of your day — and ask AI to synthesize it. Have your aha moment. Then do it again tomorrow. And the day after.
Within two weeks, you’ll understand AI better than any briefing document could teach you. You’ll see opportunities across your organization that are invisible from the sideline. And your team will notice. They always do.
The best AI strategy doesn’t start with technology. It starts with you.
Take our AI Maturity Quiz to see where your organization stands today, or book a call to discuss what an executive AI briefing looks like for your team.