Friday, September 11, 2026
FRIDAY – AI FOR THE C SUITE®
Read time: 8–9 min · Read online
Hi, it’s Chad. Every Friday, I serve as your AI guide to help you navigate a rapidly evolving landscape, discern signals from noise and transform cutting-edge insights into practical leadership wisdom. Here’s what you need to know:
1. Sound Waves: Podcast Highlights
This past Monday, I was joined by Ross Barnes, founder of Galahad and the man who’s taught AI literacy to more than 1,000 senior leaders through Emeritus alongside MIT and Stanford faculty. We discussed a story that should make every CFO and COO uneasy: the time Ross sat in an NDA’d meeting with a multinational, opened a vanilla ChatGPT account, asked about their confidential product launch, and got everything. The lesson isn’t that Ross is clever. It’s that your confidential information may already be one prompt away from anyone who asks. To hear more, tune in wherever you get your pod on. Hit one of the below links to check it out:
Apple · Spotify · iHeart · Amazon · YouTube
Subscribe for free today on your listening platform of choice to ensure you never miss a beat.
2. Algorithmic Musings: Things Are About to Get Weird(er)
I believe that we’ve crossed a point where our horizon of relative certainty will continually contract, while our relative uncertainty (that which is unexpected, unpredictable, and unforeseen) will expand. Dramatically.
In short, I think things are about to get weird(er).
I recognize this is an entirely subjective vantage point and a less-than-objective way to open an article. You might even be forgiven for suspecting this is some rough-handed attempt to clickbait you. It’s not. Instead, I feel compelled to acknowledge that sometimes the combined weight of all the tea leaves defies precise analysis and devolves into feeling.
So let’s read a few of those leaves together. I’ve picked three, and each one tells a different part of the story: capability, threat, and opportunity.
Leaf One: Capability
The debate over whether Astra and Fable qualify as AGI (Artificial General Intelligence) is getting louder. AGI remains an extremely ill-defined term, and I’m not going to wade into that swamp here. The more useful question for you and me is whether the machine can do your expert’s job, and the answer keeps coming back as “yes, and more of it every quarter.”
Consider GDPval, a benchmark built from real deliverables (legal memos, accounting spreadsheets, sales decks, engineering diagrams) across 44 occupations and graded blind by professionals averaging 14 years in their fields. In the fall of 2025, the best model matched or beat those experts on just under half of the tasks. By December the number was north of 70%, and we’ve seen several model generations since then.
Call it whatever you like. I’m comfortable calling it a problem for anyone who bills by the hour.
Leaf Two: Threat
One of the best open models has now been jailbroken. Somebody surgically removed its guardrails and put the result behind an API that anyone can rent for five dollars per million tokens. This means that cheap, capable models can now write proof-of-concept exploits that let small security shops do what only the big consultancies could do a year ago.
The flip side is that you’re now a target. An attacker doesn’t need to be skilled anymore, just funded to the tune of a few dollars. In my opinion, that makes mid-market companies with thin IT departments the softest targets on the board. You have enough revenue to be worth the trouble and not enough staff to notice the trouble coming.
Leaf Three: Opportunity
According to payment card processor Stripe, nearly 40% of new businesses on their platform in 2026 are in metros with fewer than one million people, up roughly ten percentage points from 2022.
Why does that belong in an article about things getting weird? Because for most of my career, serious new companies formed where the talent and the capital lived: New York, Boston, the Bay Area, Austin. AI is loosening that connection. A two-person shop in Lancaster, Chattanooga, or Boise can now rent something close to a full bench of analysts, developers, and marketers for the price of a streaming subscription. Your next best partner, or your next most dangerous competitor, may be forming this week in a town you’ve never scouted.
So, What’s a CEO to Do?
My argument isn’t that you need a better forecast. Forecasting is exactly the muscle that’s losing value.
Think about the storm chasers in Twister (1996, and yes, I’m dating myself again). Helen Hunt and Bill Paxton don’t win by predicting where the tornado will touch down. Nobody can. They win by getting close enough to put instruments into the funnel so they can learn faster than the other guys. Their edge comes from proximity and speed of learning, and that’s the posture I’d recommend for you, too. You need a better set of sensors and a shorter loop between what you sense and what you do. Practically speaking, that maps back onto our three leaves:
- Capability: raise your floor. Stop debating whether AI is “ready.” Pick the three most expensive knowledge tasks in your business and run them through a frontier model this month, side by side with your experts. One afternoon of that will teach you more than a year of vendor demos.
- Threat: assume you’re already in the crosshairs. Multi-factor authentication everywhere, patched systems, tested backups, and a phishing drill that includes the C-suite. Boring stuff, I know, and all of it is cheap compared to a week of downtime. If nobody in your building loses sleep over this, hire or rent someone who will. Now.
- Opportunity: widen your scouting radius. Where are the new entrants in your industry actually forming? Who’s selling into your customers from a zip code you’ve never heard of? Put someone (or some AI) on it. The competitor you can’t see is the one that hurts.
You’ll notice that none of these steps require you to know what happens next. They only require you to shorten the distance between noticing and acting.
The Weird Turn Pro
Hunter S. Thompson famously observed that when the going gets weird, the weird turn pro. With this in mind, I suggest two additional actions. First, it’s time to bring the eccentrics out from their hiding places in your organization and lean into their unique skills. Second, I suggest that when the going gets weird, the prepared turn pro. And preparation in this environment has very little to do with a thicker strategic plan. It looks like an organization that senses sooner, decides faster, and fails cheaper than the one down the road.
I recognize you’re busy running your operation. But if you’d like to talk through what your sensor array might look like, or you simply want to compare tea leaves, give me a call. Things are about to get weird(er), and I’d rather we chase this storm together.
3. Research Roundup: What the Data Tells Us
AI SEARCH SUMMARIES: THE FREE TRAFFIC CHANNEL JUST REPRICED
For a decade, organic search was the acquisition line nobody budgeted for. University of Washington researchers just put a number on what happens when the results page answers the question itself.
The numbers that matter: After AI Overviews became the U.S. default, English Wikipedia search referrals fell 5.45 percent against matched German articles and 4.82 percent against French, across half a million article pairs. About 100 million fewer search visits per month. Against Japanese, over a shorter window, the drop was 16.53 percent. For a comparable ad-supported publisher, the authors value that traffic at $10.8 million to $37.1 million a year.
What this means for your Monday morning: Sort your content into two piles. Anything that answers a factual question inside a paragraph is what an answer engine reproduces best. Anything requiring a tool, proprietary data, a transaction, or a relationship cannot be summarized out of existence. Move budget toward the second pile.
The catch: This is a floor, not a ceiling. AI summaries hit roughly 18 percent of Google searches during the study window, and only part of Wikipedia’s readers sit in the treated market, which drags the estimate toward zero. It is also an unrefereed working paper.
Action item: Ask whoever owns your analytics what search-referred traffic has done over the past 18 months, separated from direct. If nobody can answer this week, that gap is the real problem.
Read our full analysis of this research at AI for the C Suite®.
4. Radar Hits: What’s Worth Your Attention
Corporate America is getting hooked on open-source AI. AT&T doubled its open-model share since May, from 20% to 40% of its AI work, and says it’s cutting AI costs by up to 80%. Airbnb and Deloitte are shifting too. The split: premium closed models for heavy lifting, open models for simple, specialized tasks. Ask your team which workloads are paying Maserati prices for Mazda work.
OpenAI’s new reasoning technique alarms AI safety experts. OpenAI’s upcoming Astra model reportedly uses a technique that makes reasoning harder to inspect. Use appears limited, and OpenAI calls readable reasoning a core goal, but Anthropic and Google DeepMind are reportedly discussing the technique too. Running AI agents? Reasoning logs are part of your audit trail. Ask every vendor whether you’ll still see how their model reaches decisions.
China is quietly bracing for AI’s social costs. Some Chinese firms are shrinking headcount without announcing layoffs: attrition, contractor cuts, fewer graduate hires. Beijing ordered an AI employment-impact assessment system, and courts there have sided with workers replaced by AI. Turning AI productivity into fewer jobs is becoming a reputational question as much as a financial one. Decide your playbook, including where future managers come from if you stop hiring graduates.
5. Elevate Your Leadership with AI for the C Suite®
When the going gets weird, the prepared turn pro. If you want help building an organization that senses sooner, decides faster, and fails cheaper, I’m booking mid-market leadership engagements for Q1 2027 now. Reply to this email or reach me at chad@chadharvey.com. And if you know an executive who’s still waiting for AI to be “ready,” send them AI for the C Suite®. They’ll thank you before you’re finished reading next week’s issue.
Stay safe. Stay healthy. Be strong. Lead well.
Chad