Informed i’s Weekly Business Insights

THE NEWS is full of how an ignominious peace deal with Iran exemplifies a decline in American power. That conclusion could hardly be more wrong. On June 12th the Trump administration ordered Anthropic to block foreigners from Fable and Mythos, its latest and most capable frontier AI models. In an instant, everyone learned that the American government can decide who may use the world’s most important technology. You don’t get much more powerful than that.

The administration was responding to a supposed jailbreak for Fable, meaning a prompt that circumvents defences against uses such as hacking computers or making bioweapons. The chances are that it wanted Anthropic to switch off the models for everyone, and that targeting foreigners was a means to an end. Sure enough, that is what Anthropic did, while claiming that the concern about its model was overblown. The legal basis of the order remains unclear, and the ban seems unlikely to last.

What matters, though, is the demonstration that global access to the best AI may come down to a decision in the Oval Office. The administration showed in March that it is prepared to trample on the frontier AI companies, when it designated Anthropic a “supply-chain risk”. Now it has shown that it is prepared to trample on users, too.

America must decide how to wield this vast new power. The rest of the world must decide what to do about it. Even as it plans for an unreliable America in everything from defence to trade, it now has to cope with a new way of being captive to the world’s biggest economy.

This is not the first time America has tried to restrict access to frontier technologies. After the second world war it stopped helping Britain’s nuclear-weapons programme. When modern cryptography emerged in the 1970s, it blocked exports, before accepting the trade-off between having secure allies and using secrets to boost its own offensive capabilities. Uncle Sam still refuses to share its best military equipment, even with close allies. America kept the F-22 fighter for itself; allies got the F-35.

To control access to a technology, though, depends on its nature, and rivals’ ability to develop it independently. Nuclear co-operation with Britain resumed in the 1950s after it developed technology of its own; with other countries, America used the Nuclear Non-Proliferation Treaty of 1968. Cryptography methods could not be contained and eventually went public. Many countries are capable of cyber-attacks.

Frontier AI has echoes of all these examples. If the very best models can disable crucial infrastructure or help users create pandemic-ready pathogens then, like nuclear weapons, they are too dangerous for public hands. But as with cryptography algorithms, it will be hard to be sure that advanced proprietary capabilities will never be copied. Open-weight models, which anyone can download, could advance and proliferate. In cyber-security a small imbalance can bring big advantages: if an attacker has version 5 while the defender is stuck with version 4, and the better model uncovers just one more vulnerability, the weaker party will be compromised. As AI is embedded in military hardware, a similar logic may apply on the battlefield.

Yet America has a huge economic interest in leading in AI and selling its tech to foreigners. Many Anthropic staff are not American and so were hit by the ban; to freeze AI research at America’s best lab would be self-defeating. The firm also says that 80% of its consumer use is overseas. As American technology has boomed over the past decade, Europe’s payments to America for intellectual-property products have risen fivefold. America should not want to give the rest of the world a reason to team up with China, the second-ranking AI power.

Over the past several months, the authors have met with the tech and business leaders at 15 AI-centric companies—spanning continents, industries, and stages of development, from four-person start-ups to established global platforms—to learn what it takes to make AI capabilities truly deliver. The authors expected to hear 15 different stories. Instead, this diverse group of businesses, independent of one another, seemed to converge on the same fundamental insights about whatare the ground-level practices that differentiate winning companies from those that continue to struggle to get real results from their AI efforts.  The authors boiled down what they learned from these leaders into seven essential truths—hard-earned insights that collectively constitute an operating system for getting the most out of AI.

1. AI is not a tool, it’s a teammate.  The real value of AI isn’t doing the same work faster. It’s the ability to amplify the efforts of individuals with agents that function as genuine team members.
2. Know what to build and what to buy.  Build only what makes you truly distinctive. As for everything else, how far you go is a function of your own comfort level.
3. Your model isn’t the bottleneck—accessing your tribal knowledge is.  Many teams focus on which AI model to run. The ones pulling ahead focus on what their agents can find, and they invest in the knowledge layer that makes the difference.
4. Design for the swap, not the stack.  The winning architecture is not a monolithic platform. It is a thin governance layer that connects best-in-class components and keeps them interchangeable.
5. Trust precedes autonomy.  Companies build trust in AI systems through progressive autonomy: AI generates, humans judge, and the system earns more freedom only when it deserves it.
6. Centralize the platform; decentralize the tasks.  No centralized AI department can drive transformation. What works is when platform teams govern the infrastructure and business teams solve their own problems on top of it.
7. Adoption is a flywheel, not a rollout.  Successful adoption isn’t a rollout with a deadline. It’s a flywheel with four reinforcing layers: role modeling, sharebacks, measurement, and hiring.

These seven truths are more than a list of best practices. They are an agentic system—and one that meshes with McKinsey’s Rewired playbook for AI transformation: Treating agents as teammates (Truth 1) immediately raises the question of what to build versus buy (Truth 2). Building requires getting the knowledge layer right (Truth 3), which depends on a composable, governed architecture (Truth 4). Operating safely requires trust built incrementally (Truth 5). Scaling requires the right organizational design (Truth 6). Sustaining it requires adoption as a cultural flywheel, not an IT rollout (Truth 7).

Clear decision rights are crucial for collaboration: They prevent confusion, speed up execution, and reduce conflict, especially in matrixed, cross-functional, and project-driven organizations. Tools such as RACI, RAPID, and DARE are meant to help organizations define them. Unfortunately, these tools often produce only a document detailing who will play what role in a decision. And as one manager at a global e-commerce company told one of the authors in a workshop: “Decision rights are like the position plan for a children’s soccer game—a nice plan on paper that no one understands or remembers.”

One problem is that when leaders use decision tools, they often encounter resistance and skepticism. In many years of studying how power dynamics shape interactions and decisions, and in their work advising and educating more than 100 global companies in a wide variety of industries, the authors identified four common errors that organizations make when establishing decision rights.

1. Confirming Roles Without Clarifying Goals.  When teams attempt to assign roles before goals have been carefully defined, discussions about decision rights often degenerate into ego-driven turf wars. Sometimes objectives are far too broad and not broken down into concrete steps or subgoals. That makes it impossible to allocate ownership of specific decisions and identify where collaboration is needed. Other objectives, in contrast, can be too narrow or insignificant. Particularly when goals are too broad, conflicts arise among executives over the rights for decisions about them.  Before team members focus on role assignments, they all should be able to articulate the specific, measurable, and time-bound goals and subgoals.
2. Assuming Everyone Will Adhere to the Boss’s Spreadsheet.  A common and costly mistake  is treating decision rights as a static list created by a single senior leader, captured in a spreadsheet. The assumption is that once roles are assigned and documented, people will play them. But in practice, they rarely do.  High-performing teams understand that RACI and tools like it are not ends in themselves; they’re conversation starters. They prompt team members to clarify goals, confirm responsibilities, support one another in their positions, and hold one another accountable.  The obvious solution is to cocreate RACIs rather than dictate them. Bring the people who will live with the decision into the room to debate roles and resolve tensions.
3. Misunderstanding Roles.  Teams often have differing views of the behaviors expected for each role.  The fix is straightforward: Build a simple, behavioral description of each role and institutionalize it. When people know what being accountable looks like in action (how that person gathers input, facilitates debate, makes a call, and explains it), the tool stops being theoretical. The same holds true for the other roles.
4. Getting Stuck in the Same Roles.  A final mistake occurs when people get trapped in certain roles despite the best of intentions. In some cases, senior leaders are always accountable, and the people in the layer below are always responsible. In others, people act as if they’re accountable when they’re not—something that often happens when a teammate at a lower level is accountable on paper but that person’s boss still acts like the one in charge.  The best teams are intentional about tailoring roles to the topic at hand. They don’t get mired in ingrained patterns of power or deference to the formal org chart.

To organize their teams to make sound decisions, leaders cannot just articulate roles in spreadsheets. They must establish a robust process that’s woven into work—and ensures not only that people stick to their designated roles but also that the roles themselves are continually revisited as goals evolve or friction arises. Leaders and their teams should routinely ask: Did we play our positions? Where did our assigned roles support or constrain us? Such an approach will turn decision rights from static artifacts into living systems.

After several years of AI experiments and pilot initiatives, a crucial question remains open for most companies: How much of a return — and what kinds of returns — are we getting from all of this AI investment? To many executives, AI ROI still often feels more like art than science: elusive, imprecise, and industry-dependent.

Surveys and benchmarks paint a confusing picture about current returns. Much of the guidance also remains focused on measuring inputs — encouraging organizations to invest, experiment, and build capabilities (“You should invest in …”) — rather than on outputs and how to assess impact (“Here’s how to measure results”). Today, few companies apply the same financial discipline to artificial intelligence as they would to a new factory or piece of machinery.

The authors’ interviews with more than 30 CEOs and senior leaders across various industries confirm that measuring AI ROI is anything but standard practice: Two companies making nearly identical investments may define success in entirely different ways. Yet companies that fail to identify an explicit approach to AI ROI — or that simply roll out generic AI tools and hope for productivity gains — rarely realize credible, lasting returns.  ROI measurement differs by the type of AI technology being used.  AI ROI also depends heavily on industry context.  Based on their interviews with executives, the authors identified three practical approaches to measure and manage AI ROI.

1. Function-focused approach.  Focus on one business function or a small number of functions or processes. Use tailored AI solutions and metrics.  Typical measures those can be used:  Function-specific KPIs, such as response time or error rates.  Potential pitfalls could be:  Siloed metrics and no shared view across the organization.  Organizations can start scaling metrics toward a companywide AI ROI playbook.
2. Coordinated approach.  Coordinate the deployment of broadly applicable AI tools and function-focused initiatives.  Possible metrics could be:  a mix of broad operational metrics and function-specific KPIs in selected high-impact AI initiatives.  Pitfalls organizations could face:  limited comparability and fragmented portfolio-level oversight. Organizations should apply consistent financial translation and measurement logic across all AI initiatives. 
3. Enterprise portfolio approach.  Engage in enterprisewide governance of the AI portfolio.  Organizations can use investment portfolio value, NPV/IRR, business case ROI.  Organizations should watch the risk of excessive bureaucracy that may constrain early-stage or exploratory initiatives.  They should use financial and strategic metrics. Allow early bets without full ROI measurement.

Marketing and communications teams are often responsible for shaping how a company is understood by customers, employees, investors and the broader market. But their ability to do that effectively depends heavily on the support they receive from leadership.

The most effective CEOs don’t simply approve budgets or sign off on messaging. They ensure marketing has the context and influence needed to contribute strategically. Here, members of Forbes Communications Council share the most valuable ways CEOs can support marketing and communications teams—and why their actions can make all the difference.

Define The Brand From The Top.  Steve Jobs never asked what the market wanted. He decided what Apple was. The marketing team had one job: not to invent the story, but to make it visible. That is the difference. When positioning comes from the top, it becomes culture. When it comes from a marketing department, it becomes a campaign. Conviction cannot be outsourced. 

Give Communications A Seat At The Table.  The most valuable thing a CEO can do is secure a voice and seat at the decision-making table for the comms team. That access positions comms to understand, shape and drive organizational objectives directly. Otherwise, comms operates on assumptions rather than organizational intelligence, risking missed opportunities and uncovered brand touchpoints.

Lead With Authenticity.  A CEO supports marketing and communications best by prioritizing sincerity. By being visible, accessible and willing to use their voice with intention, CEOs can create a human-centric narrative. That matters because communication is most powerful and trustworthy when it feels personal, relevant and connected to the people behind the business.

Trust Your Communications Team.  CEOs should lean into trust. Of course, this must be earned, but the CEO, much like all leaders, needs to acknowledge expertise and allow for creative communications people to create plans and ideas. It is incumbent on comms leaders to explain their concepts in a clear and concise way, but at its core lies trust from leadership. It will allow for the best ideas to come forward and create new leaders along the way.

Become The Company’s Chief Storyteller.  The CEO should ideally translate the corporate strategic narrative into an inspiring, authentic, passionate and highly personal story. Be the chief storyteller and attach it to personal charisma and uniqueness to connect to stakeholders and shareholders in deeper and more meaningful ways.

The othres are:  Invest Time And Resources In Communications.  Align Marketing And Sales Around Growth.  Put Customer Insights At The Center Of Decisions.  Stay Engaged Without Micromanaging.  Provide Clear Strategic Direction.  Personally Reinforce The Brand Narrative.  Empower Marketing Leaders To Lead.  Share The Strategic Context.  Address Difficult Issues Early.  Champion Brand Building As A Business Asset.  Include Marketing In Strategic Decisions.  Consistently Champion The Company’s Purpose.  Help Marketing Speak The Language Of Business.  Align The Organization Around The ‘Why’.  And Treat Marketing As A Strategic Function.

In the age of AI, expertise is the new trust signal.  For years, investors and the public were captivated by the myth of the visionary founder whose charisma could sell a product. Something’s changed, and founders need to know about it.  Investors today want to see a combination of charisma and credibility, with a lot more emphasis on the latter. That’s because the modern world and the technology that powers it have become exponentially more complex, and it’s beyond the ability of any one leader to know it all.

In Silicon Valley, the strength of the team — the depth of the bench — has always been important. However, now professional investors are increasingly saying that the “about us” slide isn’t enough to build their trust. They want proof that the team is worth backing. 

Why investors want to meet the team .  AI tools like ChatGPT or Claude can help entrepreneurs write business plans, build prototypes, and design good-looking pitch decks. On the other hand, AI cannot replace the breadth of human expertise and wisdom it takes to build and scale a successful company.   Investors want to meet the team in person for three reasons:  Assess their expertise.  Reduce risk.  And Evaluate the startup’s potential to grow.

Investors aren’t looking for a second presenter who’s like the founder. They’re looking for evidence that the founder has the wisdom required to assemble the right experts to turn their vision into reality.  Investors aren’t just buying your idea. They’re buying into you. One of the clearest signs of good judgment is surrounding yourself with people who know what you don’t.

The biggest mistakes Amazon sellers make when choosing products come down to a predictable set of patterns: choosing products based on personal interest rather than market data, underestimating the margin impact of Amazon’s fee structure, entering categories dominated by entrenched competitors with no clear path to buy box share and committing to inventory before validating supplier pricing at real volume.

These mistakes aren’t rare! They are the standard experience for first-time Amazon sellers, and they account for a significant portion of the stores that launch with enthusiasm and stall within six months. Understanding them in advance doesn’t make you immune, but it does make you considerably harder to surprise.

Mistake 1: Falling in love with the product.  Passion for a product is not a market research strategy. It is a bias that makes it harder to read the data clearly.  The sellers who consistently find good products approach the process like a business analyst, not a shopper. They are looking at BSR trends, review velocity, competitor pricing and margin math. They are not thinking about whether they would personally enjoy selling the thing. They are thinking about whether the numbers work.

Mistake 2: Ignoring the fee structure until it is too late.  Amazon takes a cut of everything.  Margins do not survive contact with Amazon’s fee structure unless they were calculated with Amazon’s fee structure in mind from the beginning. Model the full cost stack before you get attached to a product. The FBA revenue calculator is free. Use it early and often.

Mistake 3: Underestimating established competition.   The categories worth entering are the ones where demand is real, competition exists but has not fully consolidated, and a new seller with strong account health and competitive pricing has a genuine path to buy box share. Those categories exist. Finding them takes patience and a willingness to evaluate 50 product ideas to find five worth pursuing.

Mistake 4: Skipping supplier validation.  Find a supplier and discover the minimum order quantity is 500 units at a cost that completely breaks the margin you modeled.  Getting supplier quotes early also surfaces other variables that affect the viability of a product. Lead times affect how quickly you can restock a product that is performing well. Minimum order quantities affect how much capital you need to commit before you know whether a product works. Supplier reliability affects whether you can maintain the in-stock rate that Amazon’s algorithm rewards.

Mistake 5: Choosing products based on personal purchases.  This one is subtle, and it trips up smart people specifically.  The problem is that your purchase behavior is a sample size of one. What you want from a product, what price point you consider reasonable, what features you prioritize — none of that is market data. It is an anecdote dressed up as insight.  Amazon gives you access to actual market data. BSR shows you what people are buying right now. Review analysis shows you what buyers in a category value and complain about. Search volume tools show you what people are actually looking for. All of that is more useful than your personal purchasing preferences, no matter how confident you feel about them.

Mistake 6: Choosing trend products over evergreen ones.  Trending products are exciting. Evergreen categories, the ones with consistent, year-round demand that do not depend on a news cycle or a viral moment, are boring in the best possible way. Boring products that people buy consistently throughout the year are the foundation of a stable, scalable Amazon business. Save the trend chasing for categories where you already have established supplier relationships and can move quickly. Build the base of your product catalog on things people reliably need.

Mistake 7: Not stress-testing the numbers.  The margin math worked. The demand looks real. The competition is manageable. Everything checks out.  Now ask what happens when one thing goes wrong.  A series of ‘what if’ questions.  Products that only work when everything goes according to plan are fragile. The ones worth building a business around are the ones that still work when the plan meets reality, which it always does, and reality wins.  Stress-testing isn’t pessimism… It’s what separates the sellers who are still in business two years later from the ones with a great story about the time they tried Amazon.