The Question Nobody Opens With

I spent a week in late May with executives across South Africa: a client event first, then a string of one-on-one conversations that started in very different places and kept ending in the same one. Two questions came up in nearly every room: “Are we behind?” and “Is this actually worth what we’re spending on it?”

And then there was a third question, the one nobody opened with, but almost everyone asked once they trusted the room:

“Do we even know what the AI-native version of our business looks like?”

What I told them is that you’re in the same spot as everyone, regardless of geography or industry. The noise doesn’t make it easy. A new model drops, a headline breaks, and every leader feels obligated to do something about it that same week. The most valuable skill of a leader right now is filtering noise to find signal, then deciding what your team isn’t going to do.

In the previous issue of IQ, I argued that the center of gravity has moved: value migrated from human execution to system execution, while most org charts stood still. Since then, the ground has kept shifting.
Five assumptions leadership teams have been quietly leaning on expired this spring. Each one raises the price of standing still.

The first expired assumption: “We need to pick the right model.”
For two years, every CIO conversation included some version of “which model do we standardize on?” That question is dead, and good riddance. The frontier labs stopped behaving like model companies and started behaving like deployment companies, embedding their systems inside the tools where work actually happens and dismantling the security objections enterprises have hidden behind. The labs figured out what most of their customers haven’t: handing people a tool isn’t enough.

Value shows up only when somebody redesigns the workflow around it. If you’re still deciding which model to buy, you’re answering a question two years late. The harder one: who inside your organization is actually capable of redesigning the work?

The second: “AI costs are spiraling.”
The headlines feed this worry: pricing resetting from fixed licenses to metered consumption, compute bills that rival the payroll they were meant to augment. But two problems are being conflated. The first is deployment discipline. A new flagship ship, everyone rushes to it, and the previous model, working fine three days earlier, gets dramatically cheaper the same week.

Chasing the frontier by reflex is a choice, not a cost structure. The second problem is the real one: most organizations are scaling AI with no measurement discipline. Picture the person who consumed $10,000 of compute last month. Do you know what they produced? Can you compare them to the person who consumed $200? Almost nobody can. The story under the cost panic isn’t that AI got expensive. It’s that almost no CFO’s office can answer the only question that matters: which people are creating value, and which are just burning tokens?

The third: “The labor question can wait.” For two years, AI displacement was a bullet point on a strategy deck. This spring, it broke into courts, legislatures, and even the Vatican: a papal encyclical on AI, a foreign court ruling that AI-driven dismissals don’t qualify as valid terminations, and major banks shifting their earnings-call language from “productivity” to “labor reduction.” The cautionary tales are public, too: organizations that cut deeply while degrading the experience of the people who stayed have become the case study every Fortune 500 employee has now seen.

If reducing headcount with AI is part of your next twelve months, the operating environment just got materially harder. And there’s a second question almost nobody asks out loud: what about the people who stay? The leaders who navigate this well will answer both in the same breath.

The fourth: “Technology is a bottleneck.”
The clearest signal came from the labs themselves. When the companies building the most capable models stand up dedicated deployment businesses to compete with the major consultancies, they’re saying it out loud: the model is not the bottleneck. The last mile is.

Inside the labs, coding agents went from impressive demo to producing most of their own engineering output in roughly a year. That gap, from “the demo is amazing” to “this is how we ship work,” is the one most enterprises haven’t crossed. The capability arrived, but the operating model didn’t. The leaders crossing it have stopped framing AI as a tool rollout and started treating it as an organizational redesign: which workflows get rebuilt around digital coworkers, which roles get rewritten before they’re filled, and what people must learn to direct the work rather than do it.

The fifth: “Originality is our moat.”
This is the quietest shift and the one I’d watch most closely. AI systems have now produced novel, formally verified results in pure mathematics: open problems that stood for decades, solved for a few hundred dollars of compute, with proofs a human can audit. That is not AI helping researchers; it’s AI generating original work that researchers verify and publish. Several knowledge-work categories your business depends on, from legal opinion to regulatory strategy to R&D, price themselves on the scarcity of human originality. That assumption was just contradicted in one of the hardest domains there is. Which of your moats can a verification loop and a modest compute budget now produce?

The Question That’s Left

Can you describe the AI-native version of your business? Not the version with AI added, but the version designed around it. If you can’t, you’re in good company. But then the question becomes the one that actually separates organizations: who, specifically, is working on it? Most of the time, the honest answer is nobody yet. That’s the hard truth of transformation. The technology will keep improving whether you redesign or not. The only variable in your control is whether your organization does the describing, or gets described later as the company that had every tool and changed nothing.

Executive Perspective:
Rachel Goodger, Chief Revenue Officer, CrowdIQ

Rachel Goodger is Chief Revenue Officer at CrowdIQ, a crowd analytics platform that uses computer vision to measure real-time fan demographics and behavior at live events. Her work helps teams, venues, and sponsors understand who is in the building, how fans engage, and where commercial opportunities are being missed. She brings expertise in revenue strategy, sports partnerships, fan intelligence, AI-ready data, sponsorship optimization, and live-event commercialization at scale.

IQ: Teams today are surrounded by data and visibility into fan behavior, yet decisions don’t always change. Why is translating insight into action still so difficult?

Ms. Goodger: The surprising thing is that most teams still do not have true visibility into fan behavior and engagement. They are inundated with data, which can create confidence, but it does not always show who is actually in the building or what they are paying attention to. Ticketing data is valuable, first-party, and actionable, but it often captures only the buyer, not the full group they bring. Surveys have similar limits, because they tend to overrepresent the loudest voices. At CrowdIQ, we are focused on closing that gap by giving partners insight into every fan in the venue. The last mile starts with awareness that behavioral intelligence now exists.

IQ: Many organizations design experiences based on assumptions about their audience. Where do you see the biggest gap between who teams think is in the building and who is actually there?

Ms. Goodger: The biggest gap is in the fundamental assumption of what a sports fan looks like. If you and I think about a sports fan, we probably have a similar image in our heads. But that is not necessarily who actually shows up. What we consistently see across major leagues is that fans under 40 are almost always split about fifty-fifty male and female, which challenges a lot of traditional assumptions. There is also a perception that fandom skews older, when in reality, we are seeing strong turnout from younger audiences, including Gen Z, often in much higher numbers than expected. The issue is that these groups are not reflected in ticket buyer data. That is where the disconnect begins. We worked with an NBA partner over the course of a season, where their ticketing data showed about 4% of their audience was Gen Z. When we measured the actual crowd, it was closer to 22%. That is not an anomaly. It is something we see consistently. A lot of that comes down to who is purchasing the ticket versus who is actually attending. One person buys the ticket, but the people who show up can look very different. So you end up designing experiences, content, and partnerships around an assumed audience that is not actually there.

IQ: Sponsorship has long been sold on assumed reach rather than verified engagement. What does it take for organizations to shift from selling exposure to proving real audience value?

Ms. Goodger: In-venue sponsorship has been built on a proxy: attendance stands in for engagement, and reach stands in for attention. Historically, if 18,000 fans were at a game, that has been your impression count. But those 18,000 people are not looking at the scoreboard or signage for the entire game. CrowdIQ measures attention and engagement in real time—where fans are looking, what they are engaging with, and how that changes moment to moment. That allows you to benchmark engagement across the field of play, the big screen, activations, and fans’ phones. It creates a fundamentally different conversation with sponsors: evidence, not assumptions.

IQ: As behavioral intelligence becomes more accessible, what will separate organizations that turn it into an advantage from those that don’t?

Ms. Goodger: For me, it comes down to revenue and time. The organizations that adopt behavioral intelligence and lean into AI are going to make more money because they are saving time. It is not about replacing the workforce. It is about making everyone’s job easier and enabling them to do more.

If you can query your data and get an answer in 10 seconds instead of spending an hour manually pulling it together, that changes how quickly you can operate. It changes how decisions get made, how campaigns are executed, and how teams spend their time day to day.

That efficiency ultimately translates into revenue. It allows organizations to move faster, be more precise, and take advantage of opportunities that they might otherwise miss. And then, over time, that becomes a meaningful competitive edge.