Silvio Sangineto has some ideas about the future of AI, how it’s going to change business, and how, eventually, leaders will be able to integrate it into what companies were doing before the last decade pushed us into a new kind of computing.

First of all, AI wrote the foreword. You can check that out for yourself.

After this initial page, where the author asked the machine to tell the truth about its own limits, Sangineto took the reins, and with a preface, started to sketch out a global scenario where we have to address what he calls the Human Readiness Gap: take the bull by the horns, and harness AI in the right ways, to prosper.

“Our systems move faster than our judgment can mature,” Sangineto writes, “faster than we can decide when to deploy them, how to govern them, and who is accountable when they operate beyond our control. … Faster models. Cheaper inference. New stacks. Talent shortages. Reskilling initiatives. We debate which jobs will vanish and which industries will fracture; we interrogate the machines relentlessly, but rarely interrogate the layer of society making the decisions.”

This observation is going to frame the body of a book that relentlessly asks the relevant questions about how we as humans compete, in the AI age, not just with each other, but with the technology itself.

One of the earliest thesis statements in Sangineto’s book is the disappointing metric trotted out by McKinsey and others about the rate at which businesses are successful in their AI pilot programs.

“The models are capable,” Sangineto writes. “The infrastructure is available. The tools, at this point, are not the bottleneck. The bottleneck is human. The systems are moving faster than the people deploying them are growing.”

I liked this additional line:

“It’s called the human readiness gap. I have yet to walk into a room where it wasn’t there.”

Later, Sangineto gives us a whole lot of labeling to navigate his concepts, writing about Executive intelligence and the ‘intelligence OS,’ and human roles like the narrative shaper, systemic steward, and the curious operator, the ethically ambitious leader, the ambidextrous thinker, and the radical unlearner. Sangineto also mentions four ‘leadership taxes’ that will come into play again later.

For reference, Sangineto also gives the reader a game plan at opening, a road map to help the audience plan, including four stages: diagnosis, capability, implementation, and reckoning. The first three chapters, he lays out, reference the “decoupling” of key components, where chapters 4-10 involve the “seven mindsets” of human response. Sangineto dedicates chapters 11-13 to “the playbook” and actionable ideas, with the final chapter geared toward a “future landscape.”

“The seven mindsets are learnable,” Sangineto writes. “The four taxes are payable. The organizational redesign is achievable. The governance frameworks exist. The leaders who demonstrate these capabilities are named in these Introduction pages, their decisions documented, their trade-offs preserved. What remains is the decision to use them.”

Here’s a framing that Sangineto delivers right at the beginning of Chapter 1:

“I have never watched an AI initiative fail for lack of money. The budget arrives, usually faster than anyone expected, approved in a meeting that took twenty minutes, because nobody in this cycle wants to be the executive explaining why they didn’t. Then licenses. A rollout plan. Champions named in every function, enablement sessions on the calendar, a launch with a name and a logo. And then a number, to show it worked.”

This noted, Sangineto discriminates between the information era, and the era of intelligence, explaining that decoupling that has to occur.

The barrier’s not computational anymore, he suggests, it’s organizational.

I also wanted to include a set of caveats that Sangineto includes, explaining why “measurement theatre” can be performative, and articulating what a skeptic might say, attributing this line of thinking to an Embedded Skeptic, a character that will turn up again throughout the book.

“‘Full maturity’ is a bar somebody chose, and they chose one almost nobody clears,” the Embedded Skeptic suggests. “What they call ‘pilot programs’ we call prudent risk management. Every transformative technology—cloud, mobile, the internet itself—showed similar adoption curves. Declaring a crisis five years into the generative AI era is premature. The real irresponsibility would be betting the farm on ROI models that don’t exist yet.”

That’s a compelling argument – that we must give it time.

Later, Sangineto goes back to his four taxes, suggesting that executives need the humility to realize that they can no longer predict certain kinds of behavior, and that they need to be unlearning some of the engineering principles that no longer apply.

Here’s a citation he gave in Chapter 2 that underscores this:

“Economists Ajay Agrawal, Joshua Gans, and Avi Goldfarb argue in ‘Power and Prediction’ that AI functions primarily as a prediction machine,” Sangineto writes. “It transforms uncertainty into probability. When the cost of prediction drops to near-zero, the value of its complement—judgment—rises exponentially … before AI, network engineers managed boxes: access points, switches, routers. They predicted where failures might occur based on historical patterns and manually intervened. Prediction and remediation were human work. With AI, the system itself predicts anomalies and auto-corrects. The engineer’s role doesn’t disappear; it shifts upward.”

Standard Operating Procedure, he says, is frozen prediction.

Chapter 3 involves more of these lessons leaders must learn, and the “taxes” they must pay to learn them. With that, Sangineto moves into another section of the book, a very specific one, in which the author explores the Seven Mindsets that are the backbone of this human analysis component.

Chapters 4-10: The Seven Mindsets

Instead of a detailed report on this part of the book, I just want to go over each of Sangineto’s seven mindsets, or personas, briefly, because I found them to be really compelling.

The Curious Operator – here, Sangineto is talking about actually doing the work, getting familiar with systems, and having the curiosity to immerse yourself in what’s necessary to understand the realities that we are in.

The Radical Unlearner – Sangineto suggests we must strip away obsolete skills, and learn new ones. That can be difficult, and humbling.

The Probabilistic Strategist – I really liked this one. Let me include this definition verbatim:

Sangineto writes that the role “replaces the comfort of certainty with the discipline of calibrated uncertainty.” That says more about the technology than it does about the persona – but the goal is there, written in large print. We have to anticipate the complexity of these technologies, and the resulting decisions that we make about them.

The Narrative Shaper – this is the storyteller, and the storyteller, as Sangineto notes, must be truthful.

The Ethically Ambitious Leader – “makes trust the moat.”

The Ambidextrous Thinker – here’s the twofold goal that Sangineto invokes: first, understand what exists, and second, innovate what comes next.

The Systemic Steward – I understood this role to refer to how to handle AI in the right ways. Here’s Sangineto’s definition: “the leader who understands that the Intelligence Era does not reward isolated brilliance; it rewards those who can build systems that are wiser than any single decision.”

“These are orientations rather than a checklist,” Sangineto adds, “ways of seeing that, taken together, constitute a new kind of executive capability.”

The first five words of Sangineto’s “Playbook” section, in a sense, say it all:

“Mindsets without mechanics stay aspirational.”

The book proceeds to handle the org chart, the individual role, and the processes that take place in the boardroom, all with an eye on structure, roles, and governance.

Part of the imperative, Sangineto suggests, is a culture that will implement a framework. We all need to be “makers.”

He also cautions leaders to avoid Parmeggiani’s ivory tower, where work suffers from a particular lack of engagement. (Simone Parmeggiani is the Product Design Director at Meta’s Superintelligence Lab.)

“The most dangerous failure mode for an AI leader is not making the wrong decision,” Sangineto writes. “It is making decisions without the firsthand context to evaluate their quality.”

Another idea that emerges is that AI can, in a way, truncate the ‘synapses’ between the individual roles, flattening some processes, and streamlining others. He mentions trainers, or domain people, and explainers, who, in some way, translate, as well as sustainers, who support.

All of this, he suggests, works in concert to produce the collaborations that matter. But throughout, Sangineto keeps coming back to the necessity of bold action.

“The adoptions that work are the ones where people are brought into the decisions early and told plainly what the technology is meant to do for their role,” he writes. “That sounds obvious. It is rarely done, because doing it asks a leader to speak before they are certain—and on this subject most leaders are not certain, and they know it. So they wait for something definitive to say. The waiting is the damage.”

Check out Sangineto’s set of “Monday Morning Actions” aimed at breaking through this paralysis.

In the next chapter, Sangineto has something unique to say about job displacement:

“The question most organizations are asking: ‘Which jobs will AI replace?’ is the wrong question,” he writes. “It generates the wrong analysis, the wrong anxiety, and ultimately the wrong response. Jobs are bundles of tasks. And AI does not replace bundles. It targets individual tasks within them, specifically the ones that follow rules, recognize patterns, and process information at scale. The job, as a unit of analysis, is almost meaningless in this conversation. The task is what matters.”

It’s not tech optimism, he asserts, or tech pessimism; it’s in the middle. Regardless, Sangineto has this to say about the interplay of humans and AI:

“AI lowers the cost of prediction toward zero, which exponentially raises the value of its complement: human judgment.” As examples of human-led process, Sangineto mentions developer evolution, task-level quality, and breaking down silos, managing the people behind the tasks.

Going back to labor advocacy, here is one more piece from that chapter. This is the Embedded Skeptic speaking again:

“You are assuming everyone can transition from prediction to judgment work. That is an elitist premise. Not everyone wants to be a creative problem-solver. Many people chose their roles precisely because the work is structured, predictable, and clear. You are dressing up downsizing as empowerment. And your four-step framework assumes organizations have the time and budget for careful redesign. Most don’t. They’re running lean, under pressure, and a thoughtful task inventory is a luxury, not a plan.”

Sangineto then goes over how to deal with this argument, in detail. That’s a good read.

In the next chapter, on boardroom process, Sangineto cites the work of entrepreneur Giacomo Marini , in laying out some actionable principles of building “a governance framework built to last.”

He also lays out three “friction points” in governance frameworks, which I’ll enumerate: one, a board’s “AI fluency gap,” where top people haven’t spent enough time around this stuff, the “conformity trap” signifying less diversity in thought and approach, and three, what the author calls the “accountability diffusion problem,” to which my first thought was: “passing the buck?”

“The most pernicious governance failure is the one that looks, from the outside, like governance is working,” writes Sangineto. “Organizations with elaborate AI governance frameworks (ethics boards, model review committees, fairness audits) can still fail to produce clear accountability because accountability is distributed across so many functions that no single person or body holds it fully.”

So yeah, passing the buck.

Following this, Sangineto lays out some advice for leaders, based on the above challenges. I’m going to include all of this verbatim:

“In the C-suite, your board is not equipped to ask the right governance questions yet. Your role is to build the briefing infrastructure that makes meaningful oversight possible. That means model cards before deployment, not after. It means human-oversight tier documentation before the board approves capital expenditure. It means quarterly AI health reviews on the committee agenda before the first adverse incident forces the conversation. You are not protecting the board from governance; you are giving the board the tools to do it. As a founder, you will reach a scale where your governance architecture determines whether institutional investors, enterprise customers, and regulated-sector clients will trust you with their data and their processes. Building governance infrastructure early, when your team is ten people, not a thousand, is not overhead. It is the trust moat that allows you to deploy in domains your competitors cannot enter.”

That’s a mouthful, but it does address the accountability dilemma.

Sangineto’s Monday Morning Action, here, is a governance audit, and his recommendation is that this kind of leadership should be ongoing and incremental.

“The governance architecture this chapter describes is not built in a day,” he writes. “It is built in a sequence of deliberate choices, each one placing a named human with defined authority at the point in the decision chain where consequence can be owned. You start with one system. You add the model card. You assign the tier. You schedule the review. Then you do it again, for the next system.”

With all of this laid out, in the book’s eventual conclusion, Sangineto turns an eye to the future. Again, he adds labels, this time, names for three possible outcomes: The Bridge Built, The Bridge Burned, The Bridge Half-Finished.

This stuff is so neatly metaphorical. I love it.

I’m going to leave it to the reader to find out about these choose-your-own adventure endings. Look for terms like narrative crisis, compression paradox, and again, those leadership taxes. In a “mandate for 100 days” Sangineto provides examples of weekly goals, to get started turning ideas into solid practice. He ends with a call to action (and an epilogue) suggesting that we need to seize the moment that we are in:

“What separates technological possibility from societal trust is not a shortage of time or capability,” Sangineto concludes, “it is a shortage of nerve. Courage, it turns out, is in scarcer supply than compute.”

That’s a good ending point. Take a look as this book hits the shelves. And stay tuned.