THE GREAT AI DISRUPTION: HOW TO MANAGE in the AGE OF AI
A STRATEGIC ROADMAP
PART 1:
Overview of how AI is intersecting with and massively disrupting macro trends, economics, and geopolitics.
PART 2:
HOW TO MANAGE IN THE AGE OF AI
All this disruption creates an unprecedented set of challenges for managers and leaders, and indeed it’s evident that the very practice of “management” is already shifting in conjunction with the changing rules of the changing economic system.
Further, it’s clearly not sufficient to merely grasp the macro picture of disruption that AI is fostering, nor even to pinpoint the specific factors that are changing. We also need to understand in detail what’s happening at the organizational and operational levels in order to work out the consequences for our own organizations, and begin make the necessary adjustments.
For business organizations, the arrival, scaling, and increasing impact of AI means that …
Major shifts in customer expectations and experiences
Fundamental changes are also emerging at the most basic levels of how work is done, whether in offices, factories, or distribution centers,
Which in turn changes cost structures for business operations,
And then shifts competitive advantage in sectors and industries to AI innovators.
This situation both enables and requires workforce redesign,
All of which in turn means major changes in executive management across the firm, in strategy, and in HR
AI and other emerging technologies also alter the risks and risk exposure,
And will soon cause or require new forms of regulation.
The broad scope of this list this leaves very little out of the picture.
CHANGING HOW WORK IS DONE:
AN INVENTORY of ORGANIZATIONAL CONSEQUENCES and DISRUPTIONS
A profound restructuring of global business practices and their macroeconomic foundations is occurring. The initial era of AI experimentation has transitioned into a highly demanding period of operationalized, AI-native business execution, and over the coming two to three years, success (and possibly even survival) will be heavily influenced by how effectively organizations transition from a human-only workforce to a workforce composed of humans, plus AI, plus robots.
In addition, autonomous multi-agent networks will become prevalent, which will force organizations to change their IT compute architectures, data management and governance frameworks, and capital allocations.
In this section we’ll explore important shifts already occurring in eight significant aspects of business operations. We’ve chosen these eight because they’re at the forefront of change, and because taken together, they cover nearly the entire span of what most businesses do.
CUSTOMER EXPECTATIONS AND BUSINESS MODELS
PURCHASING AND PAYMENTS: SHIFT TO OPERATIONAL AUTONOMY FOR AI
SOFTWARE DEVELOPMENT ACCELERATION AND PRODUCT-CYCLE COMPRESSION
THE NEW INFRASTRUCTURE
ORGANIZATION, WORK, AND WORKFORCE REDESIGN
REMEMBER, IT’S NOT JUST AI: ROBOTICS AND. …
MANUFACTURING, DISTRIBUTION, AND TRANSPORTATION ARE ALL BEING TRANSFORMED.
THE NEW DEMANDS ON HUMAN RESOURCES
All of these, of course, are essential functions that enable businesses to carry on. Change one, and the organization changes. Change them all, as we are seeing now, and the word “change” is hardly adequate to describe the impact. You’ve got to start talking about massive disruption and profound transformation instead.
Further, all of them are essential components of every firm’s business model, so these changes mean new business model threats and opportunities. We use the term “business model” to describe how a firm goes to market and how it organizes the back end to make going to market possible. It’s clear that the changes being wrought by AI are altering both the front and the back ends so thoroughly that old business models are becoming obsolete exceptionally fast, while new ones are emerging everywhere.
Consequently, the threats to established firms are very real, because of the ruptures and opportunities that these new capabilities present.
In aggregate this means that literally all aspects of management, strategy, and decision making are undergoing significant change, as essentially nothing about the enterprise, inside or outside, is unaffected by the rise of AI. The impacts that AI tools and methods are already having now, and will continue to have even more powerfully as they expand their reach, will be utterly transformative.
So below we will look at each of these eight factors in more detail, and then we’ll consider some of the key strategic consequences in the section that follows.
EIGHT IMPORTANT SHIFTS IN BUSINESS OPERATIONS
1. CUSTOMER EXPECTATIONS AND BUSINESS MODELS
AI is inducing fundamental shifts in how customers behave, and in what they expect. As instant and personalized service becomes ever more possible, anything less becomes unacceptable. Faster and more precise answers, easier transactions, more personalization, and 24/7 service across all available channels are the new standards, and firms that cannot deliver high levels of AI-assisted service quality will gradually (or quickly) become less relevant.
And yet firms that over-automate may damage trust, and the fine line between too much and too little will not necessarily be easy to determine. Getting it right will become a source of competitive advantage.
And so will shifting the firm’s business model to both meet new customer expectations and take advantage of new capabilities that AI provides.
The nuances around business models are so important in these times of advancing technologies and accelerating change that we have written on them extensively, including an entire book, Business Model Warfare, and a couple of white papers that are being used in business schools worldwide.
The gist of these writings is that innovation and change create countless new possibilities for business model innovations that can enable powerful new entrants to the market, and many of these innovations are severely disruptive to established firms. However, many business leaders simply take their existing business model as a given, and fail to think creatively about how change could be leveraged to improve their model, and how it also presents a threat.
The long and continuing list of firms that go out of business because they have failed to adapt their business models includes many well-known failure stories. The list includes Nokia, Kodak, ATT, Sears, Kmart, Oldsmobile, and hundreds of others.
Conventional approaches and sound management practices aren’t enough. A very famous and popular business book entitled Good to Great reported on eleven companies that in the author’s opinion exemplified the world’s best management practices. Within a decade, three had failed outright, four more were struggling, and only four were still leading their sectors. What did the failures and laggards in? Obsolete business models that became irrelevant in the face of massive change.
Conversely, here are two examples of how leading firms innovate their business models, Nvidia using a shift in how it sells, and Walmart using AI to change how it manages inventory and supply chain.
Nvidia is letting AI startups skip the hardware bill for a share of the profits
AI startups can now access Nvidia's computing infrastructure without paying for the hardware upfront. Rather than finance costly data center buildouts, startups can pay Nvidia a share of the revenue they generate using its GPU capacity, shifting Nvidia to a business model of recurring and usage-linked revenue streams.
The arrangement removes several barriers to entry for emerging AI companies, including site selection, power procurement, construction, and hardware setup.
Nvidia framed the program as a response to a shift from building AI models to running them continuously at scale. Launch partners Sharon AI and Firmus Technologies are building data centers that will rent computing power to smaller AI companies under the revenue-sharing agreement. The move comes as Alphabet and Amazon expand access to their own custom chips, pulling at the customer base Nvidia has long dominated. Nvidia’s response is to innovate its business model.¹
AI at Walmart
Walmart has deliberately moved past exploratory AI pilots by embedding enterprise-wide machine learning and agentic AI directly into its core inventory management and supply chain logistics systems. Instead of leaving AI tools to localized store choices, executive leadership deployed centralized platforms that orchestrate demand sensing, dynamic pricing, and autonomous warehouse routing on an industrial scale. This top-down coordination has fundamentally transformed how product flows through their global network, driving multi-billion dollar efficiencies that stand out sharply against competitors who remain stuck in isolated software pilots.
As a force that drives business model disruption, these two examples show how AI enables companies to enter adjacent markets faster by automating expertise, service delivery, content generation, analytics, customer support, and logistics. Traditional advantages based on scale, headcount, geographic reach, or specialized knowledge are becoming less important, or even entirely unimportant.
In addition, some platform-based and AI-native competitors attack profitable niches before incumbents even notice, taking market share and establishing their own brand credibility with astonishing speed in an environment where customers initiate viral messages that spread almost instantaneously. And once your brand and its business model are tarnished, it’s a very long road back.
So you need to ask yourself, How is AI changing the behaviors and expectations of my customers today, how might it change them tomorrow? And how can I leverage AI to transform my firm’s business model?
A business model expresses how a firm is viewed from outside; the changes going on inside because of AI are massive as well, as we see below.
2. PURCHASING AND PAYMENTS: SHIFT TO OPERATIONAL AUTONOMY FOR AI
The consequences inside organizations reach across all facets of operations.
For example, AI agents are advancing from being mere tools in the organization’s knowledge repertoire to becoming autonomous systems that manage and execute complex workflows for the organization.
The impacts will only increase, resulting in major disruptions to back-office operations, software development, R&D, manufacturing, customer service, finance, procurement, and compliance processes. Essentially no aspect of business operations will be untouched.
For example, the traditional mechanisms of business-to-business purchasing are undergoing an abrupt and complete restructuring. By 2028, as much as 90% of the $15 trillion in worldwide B2B transactions will be directly administered by AI agents via autonomous machine exchanges acting on behalf of corporate buyers, which will of course fundamentally alter all facets of finance and accounting operations.
Already two-thirds of B2B buyers polled have explicitly said they prefer a representative-free purchasing experience, and almost all are already utilizing AI to gather information and evaluate vendors. Rather tasking humans to navigate supplier websites, download spec sheets, and negotiate with account executives, autonomous procurement agents perform vendor shortlisting, compliance verification, and contract evaluation without direct human intervention.
Clearly this is the future, which means not only major change in finance, but also that ,conventional search engine optimization and pay-per-click advertising are becoming obsolete. When the primary buyer is an autonomous software agent rather than a human, the medium of persuasion shifts from narrative branding to structured, machine-readable data.
To succeed in this transition, enterprise marketing must shift toward Answer Engine Optimization (AEO), a new marketing discipline (and a new buzz word) that’s been launched at scale by platforms like HubSpot.
A vulnerability of this approach is that machine agents don’t contact sales representatives to clarify missing or ambiguous product specifications, so if an AI agent cannot find exact, machine-readable specifications, it simply excludes that supplier from the transaction matrix. Incomplete product catalogs thus become a direct revenue leak, so with the transition to automated purchasing, firms must update their catalogs to standardized, machine-readable databases with 99.9% attribute completion. Such “golden records” see 3x to 4x higher visibility in AI recommendations, and so become yet another aspect of business that AI is affecting very directly.
In addition, commerce engines Adobe and Shopify have already integrated Agentic Commerce Protocols (ACP) and Universal Commerce Protocols (UCP) to allow machine agents to dynamically query not just product catalogs, but product inventories. This enables real-time, agent-to-agent negotiations, where seller-side agents dynamically generate optimized pricing counteroffers to match the compliance parameters of buyer-side procurement agents.
Traffic referred by AI systems now converts at 4x to 5x higher rates because these buyers are much further along the evaluation funnel, just one more example of what the shift to AI-enabled operations will mean across the enterprise.
Another consequence is in payments, as the spread of autonomous AI agents capable of executing multi-step tasks across the internet has created a financial bottleneck, simply because agents cannot complete transactions using traditional human-centric payment systems.
Hence, we are undergoing a transition from human-initiated “click-to-pay” transactions to autonomous “decide-to-pay” systems that require settlement protocols that do not depend on human identity, physical location, or traditional banks. Due to their programmability, instant settlement, and permissionless custody, stablecoins have emerged as the new native money layer for software agents. Industry analysts thus forecast that stablecoin supply will grow significantly, passing $420 billion in 2026 and continuing to expand.
Standardization is occurring around a series of innovations across all facets of the payment infrastructure. Google has created an “Agents to Payments” (AP2) protocol which allows “Human Not Present” payments using cryptographically signed mandates to create non-repudiable audit trails, a new infrastructure for autonomous agent spending.
Google donated the A2P protocol to the FIDO Alliance in April 2026. FIDO, the Fast IDentity Online Alliance, is an open industry consortium founded by tech leaders including Google, Microsoft, and PayPal that is intended to eliminate the world's over-reliance on passwords by developing secure, phishing-resistant, and interoperable authentication standards.
Take another look at the new terminology mentioned in the last few paragraphs that AI-enabled purchasing has necessitated:
Representative-free purchasing experience
Answer engine optimization (AEO)
Golden records
Agentic commerce protocols (ACP)
Universal commerce protocols (UCP)
“Decide-to-pay”
“Agents to payments” (AP2) protocol
“Human not present” payments.
These are innovations induced by other innovations, which gives us a sense of how much and how fast AI is changing just this one aspect of commerce. Across all dimensions of AI’s impact we are finding hundreds of new terms, concepts, and business practices, some of which will undoubtedly affect your own firm in fundamental ways. Just keeping up with the terminology is a serious task; keeping up with the operational requirements for change is a major one, and it’s not going to get any easier.
As these shifts and innovations create new possibilities as they also solve new problems that enable greater reach for the agentic workforce and agentic commerce. Clearly they also create new demands on organizations for new operational models, protocols, and structures, new approaches to governance, and perhaps most of all the willingness to shift how firms are managed. This is not the internet of the past, not past ways of doing business, and not an easy territory to navigate.
3. SOFTWARE DEVELOPMENT ACCELERATION AND PRODUCT-CYCLE COMPRESSION
The development and spread of AI means an even greater focus than before on the core languages in which AI also functions, which is of course software. And as we have noted, AI has already proven to be exceptionally proficient at writing good quality code. As a result, AI-powered code completion, generation, and review tools are reshaping how software gets built.
Using AI, software coding, testing, documentation, prototyping, analytics, and product iteration are all speeding up significantly. The fastest firms release, test, and revise software offerings much more rapidly in the quest for competitive advantage, and consequently the global market for AI coding assistants reached $12.8 billion in 2026, and could hit $30.1 billion or more by 2032.
About 85% of developers now use AI coding tools, but they’re not just using one, as “tool stacking” has become a standard practice. Roughly 70% of engineers are using 2, 3 or even 4 AI tools simultaneously.
All this agent-generated code induces significant changes in IT operations:
Autonomy: Among those who remain, a significant part of the effort is no longer focused on manual syntax writing, the coding itself, and moving instead toward defining highly precise specifications, detailing acceptance criteria, and determining the necessary upfront parameters. As more code is written by AI, human teams are thus shifting their efforts from being “code authors” to becoming “teams of orchestrators” of code written by AI. This changes both the work, the workforce, and the workflow.
Workforce: Large scale layoffs are now occurring across IT organizations, while hiring for people who write code has reached an abrupt dead end. Roughly half a million IT jobs were eliminated in 2025 and the first half of 2026, a significant portion of which (but not all) were directly attributable to AI. Layoffs in IT are now so common that you can follow the news at a website dedicated to the topic, https://layoffs.fyi/2026-layoffs/, which says: “Layoffs.fyi is the most comprehensive public database of tech layoffs.” SO it’s a thing, and it will continue to be a thing for the foreseeable future.
Integration: The speed at which AI can create new code has resulted two new integration bottlenecks. First, whereas the generation of the actual code used to be the limiting factor in IT operations, AI writes and tests new code so fast that the problem area has migrated downstream to QA, and to the integration of new code written by AI into deployment in operations.
Exploiting AI thus requires a transformation within the IT organization and its operational workflows. The traditional siloes dividing software developers, data scientists, and infrastructure engineers are shifting to unified practices like MLOps (Machine Learning Operations) in which IT teams shift their focus from maintaining system uptime to orchestrating continuous integration and continuous deployment (CI/CD) pipelines for AI models. This demands new skills, including AI infrastructure architects, data engineers who maintain clean data pipelines, and security specialists dedicated to safeguarding data privacy and mitigating the unique vulnerabilities of machine learning models.
Second, deployment also means not only the software integration, but the operational integration. That is, organizations can only take so many new procedures, protocols, and operations changes at a time, so code production has now surpassed the capacity of many organizations to make use of what AI can do. If a key limiting factor is the capacity of organizations to make use of the new code that AI-powered IT groups produce, then it becomes an additional responsibility of the IT organization itself to support the rest of the organization by accelerating the rate of adoption. These are critically important jobs for humans that reach into all facets of the firm, from the small operational details in front and back offices, to strategy and business model design.
All these are essential competitive factors going forward, and they all present new costs. To take advantage of these new capabilities and scale the use agents successfully, software engineering groups are finding that they must address many structural issues in how they operate.
Governance: Branch protection, code review policies, and role-based access controls are essential to define a clear chain of permission, while avoiding ad-hoc structures that are prone to errors.
Observability: Comprehensive logs must trace agent decisions, ensuring that any failures can be identified, debugged, and audited.
Evaluation: As integration is a key barrier to scaling the use of AI-generated code in IT operations, firms are developing automated, continuous measurement tools such as unit tests, linters, and QA suites that address and mitigate reliability concerns at a high rate.
Cost Controls: A fundamental tension is the motivation of AI platform companies to increase the cost of access to their tools, while organizations are wary of token bloat and runaway token expenses. Techniques such as task-based routing and the use of lightweight models for boilerplate code generation reserves expensive reasoning models for major architectural decisions, all intended to prevent AI and cloud spend from ballooning.
The combined result of all these factors has led to new approaches not just to code generation, but to the management of the IT effort and the process of leveraging IT to create competitive advantage. Organizations that operate on the leading edge of these shifts have positioned themselves to gain significant advantages, while the laggards are certain to suffer. Hence, yet another form of the AI-induced arms race is upon us.
4. THE NEW INFRASTRUCTURE
From a broad historical perspective, each new generation of compute architecture going back many decades has delivered increasing performance at considerably lower per-unit compute cost. But each shift, from mainframes to minis to PCs to cloud, and now to AI, has increased the organization’s reliance on digitalization, and while the performance of each of the essential elements of IT infrastructure has significantly improved, all this is happening at an increasing capital cost, and bringing a higher complexity cost as well.
To fully utilize the power of artificial intelligence, firms are finding that they have to overhaul their hardware foundations, shifting from traditional enterprise servers to specialized infrastructure. Standard CPUs are no longer sufficient, as organizations now require clusters of high-performance Graphics Processing Units (GPUs) or Tensor Processing Units (TPUs) capable of handling massive parallel processing workloads.
Alongside computing power, data storage solutions must evolve to provide ultra-low latency and massive throughput, as AI models need constant access to vast datasets to train and run inference efficiently. This physical architecture is often deployed via a hybrid cloud strategy, balancing intense, on-premise computing clusters for sensitive proprietary data with the elastic, scalable power of public cloud providers.
They also must modernize their networking infrastructure to prevent data bottlenecks. Traditional enterprise networks are ill-equipped to handle the “East-West” traffic patterns—data moving rapidly between servers rather than just from a server to a user—that define AI workloads. Companies are increasingly adopting high-bandwidth architectures utilizing InfiniBand or specialized Ultra Ethernet protocols, which offer the massive throughput and the sub-microsecond latency required for thousands of processing cores to communicate simultaneously. Without this high-speed interconnectivity, expensive GPUs sit idle, waiting for data to arrive, which severely reduces operational efficiency and spikes project costs.
All this has costs, and thus capital allocation is shifting toward the unavoidable need for upgrades in raw compute power, automation, and AI infrastructure, plus cloud, chips, data platforms, model access, AI security, and workflow automation. None of this comes cheap.
5. ORGANIZATION, WORK, AND WORKFORCE REDESIGN
The way that work is done across all facets of business operations is changing, and changing significantly, and it’s not just in IT. Analytical, administrative, support, and knowledge-work tasks are being automated, so businesses need less and less people to complete many internal workflows. Conversely, more people will be needed to supervise AI, judge outputs, manage exceptions, and redesign processes to leverage the new automation possibilities.
And the burdens on senor leaders and business strategists are also increasing significantly due to the rapid acceleration of AI adoption in conjunction with a world that is also changing fast due to many complicating factors beyond AI. The consequences that reach across all facets of business design and operations.
Fewer people but more AI changes the operations cost structure. AI is not free, and as noted above, organizations already see steep increases in infrastructure costs for the necessary compute infrastructure. In addition, complex multi-agent workflows are composed of countless decision steps, validation loops, and autonomous negotiations in multi-agent systems, all of which can bloat token usage. Indeed, by some estimates token consumption could grow by 3,400% by 2030, with commensurate impact on operating budgets even if per-unit token costs go down. This is accelerating the rise of on-premises “deskside agentic AI” and high-performance AI-native PCs that handle local transactional queries rather than using centralized public cloud services, but still the costs will grow.
As the investment formulas to support ongoing operations are shifting, so the development of the AI-enabled workforce is also a new category of investment that will absorb considerable capital and operating expense, yet another dimension of the pivot to full and effective use of AI across the enterprise.
With each advance, key functions across the enterprise have had to be re-thought and reconfigured, budgeting and staffing shifted, assets and risks modeled. Some of the past shifts have been relatively quick and painless, but for most organizations the shift to AI is not one of those.
AI is accelerating the breakdown of functional silos, while shifting the role of leaders from “reviewing backward-looking dashboards” to “interacting with real-time decision engines” that can automatically trigger actions based on live inputs. This means increased autonomy for AI-enabled organizational networks as traditional physical shared services centers are replaced by virtual, AI-first centers designed to orchestrate work seamlessly between human employees and autonomous agent teammates. Virtual hubs enable end-to-end automation and insight at scale, but they only work when the design of the organization is re-thought at a fundamental level, as we see here with the example of Siemens.
Siemens: New Workflows
Siemens is pioneering new industrial and software workflows by actively integrating digital agents into its engineering and manufacturing operations. Rather than using automation as a simple replacement tool, the company has restructured entire product design and factory layout processes so that engineers collaborate directly with autonomous digital twins and layout agents. Humans set the high-level boundary parameters and strategic goals, while AI agents rapidly iterate and optimize the micro-mechanics of production lines, a deep integration where human judgment and machine autonomy complement one another.
All of this certainly does have major implications for, as we said above, the design of the work, the design of the workforce, and the business model through which the organization goes to market. To further complicate the problem, no matter now good the solutions and innovations the firm invents and implements today, by next year the whole thing may have to be done all over again. The macro process of change is not a static shift from state A to state B, but rather a continuous progression from state A possibly all the way to Z.
So the problem of organization design not a once-and-done situation, it’s an ongoing challenge. Which, of course, reflects the fact that the development of AI itself is a continuous progression toward a deeply unknown state but which could soon be AGI, with all its massive consequences.
6. REMEMBER, IT’S NOT JUST AI: ROBOTICS AND. …
Artificial intelligence is moving beyond the digital screen to directly interface with physical infrastructure, a trend known technically as Physical AI, or more simply, robotics, smart equipment, and drones.
The effective deployment of Physical AI requires still more infrastructure investment, as organizations have to create secure, federated data ecosystems that allow manufacturers to share operational and IoT data for collaborative analytics and cross-company AI training without transferring data ownership outside regional jurisdictions.
This is highly visible in the transport sector, where nations are building sovereign mobility clouds to retain vehicle telemetry, high-definition mapping, and autonomous driving data within national borders, ensuring that sensitive real-time transit intelligence remains protected.
And as we noted in Part 1, AI is not the only technology that’s causing deep and wide changes across science, engineering, and operations. Quantum computing is one of the most powerful, no longer a distant theoretical project but one that is even now crossing the threshold to effective commercial use.
For example, quantum computing is already enabling breakthroughs in asset-heavy, research-intensive sectors such as chemical and materials industries, where quantum algorithms are substantially improving computational capabilities, allowing companies to model complex molecules and optimize chemical reactions in fraction of the time it used to take. New vaccines, new drugs, new materials, all enable new possibilities (and disrupt incumbents).
In finance, quantum computing is also disruptive as it is deployed to run complex portfolio optimization, multi-variable risk simulations, and supply chain logistics mapping.
Similarly, the ramifications reach across all facets of manufacturing and distribution.
7. MANUFACTURING, DISTRIBUTION, AND TRANSPORTATION ARE ALL BEING TRANSFORMED.
How we make stuff and how we deliver are all being transformed. The fifth book in our AI series is titled Post 2 Platform, and describes the transformation of the global postal industry from its historical origins as a set of national networks for delivering letters to an integrated, AI-enabled global systems for delivering anything, from anywhere, to anyone.
The book explains how AI is now and soon will be used across all facets of postal network operations, from autonomous vehicles to digital twins to real-time adaptation to bring the posts into the 21st century as powerful participants in the global logistics market. (The book is a communications tool for the consortium of postal operators that our sister firm, FutureLab, has organized to co-develop a comprehensive AI infrastructure for posts worldwide.)
A good picture of how AI is transforming industrial manufacturing is provided by taking a look at Michelin, which we did by interviewing Yves Caseau, Michelin’s Group Chief Digital and Information Officer.²
At Michelin, artificial intelligence has transitioned from a localized experimental tool to a core pillar of global manufacturing operations across its 100+ factories. Under the guidance of Caseau, the tire manufacturer leverages AI to absorb the deep complexities of tire production, optimizing everything from electricity consumption to raw material waste. By combining classical operations research with modern digital twins, Michelin simulates factory processes in real time to refine efficiency before physical production even begins. These systematic improvements across the supply chain and shop floors have already delivered an impressive €50 million in annual savings, with Caseau setting a strategic target to scale those savings to €500 million by 2030.
A significant portion of this manufacturing transformation relies on equipping tens of thousands of factory operators with specialized AI tools, emphasizing a "human-in-the-loop" philosophy. For example, Michelin relies heavily on deep learning neural networks for advanced computer vision, automating meticulous quality control checks to spot tire defects that might elude the human eye. For maintenance, the company has rapidly deployed generative AI assistants to help on-the-ground technicians troubleshoot complex machinery failures instantly. By transforming historical documentation into accessible, interactive dialogues, these AI tools enable factory teams to solve maintenance bottlenecks quickly, minimizing costly downtime while safeguarding the company’s specialized industrial expertise.
Michelin’s profits in 2025 totaled €1.6 billion on sales of €26 billion. The €500 million that Caseau is targeting will thus constitute an enormous contribution to the firm. Since this is not a secret in the tire industry, all the major firms have been forced to follow Michelin’s lead, and thus the entire industry is undergoing a massive transformation.
Nor is this a secret in any other industry, and thus they’re all obliged to undertake similar initiatives in order to remain competitive. As a result, the entire global manufacturing sector will look quite different by 2028 than it does today.
This is very nice news for the leading AI, hardware, cloud computing, and consulting firms, because they’ll sell a tremendous amount of tokens, hardware, and services to make this global transformation possible. Indeed, the scope of change suggested here, and the cost of doing it, and the returns to be earned, are a major part of the justification for the enormous valuations that AI providers are getting.
If Caseau is correct in his targeting, and Michelin really can increase its corporate profits by 30% due largely (but not exclusively) to AI, then industry as a whole will literally have no choice but to follow along.
8. THE NEW ROLES FOR HUMAN RESOURCES
If a firm doesn’t have people with the right skills, it either has to teach them, or hire, or both. Because working with AI is so different than what’s come before, human resources leaders are dealing with an entirely new set of personnel issues, which includes now only how people are trained, but what they are trained in, what the work is, and even how the work itself is organized.
The impact of AI on the workforce is creating a talent problem, a massive AI skills gap, which is forcing firms to prioritize educational initiatives to raise baseline AI literacy. By 2027, nearly all hiring processes are likely to include standardized testing or certification for AI proficiency because basic AI fluency is soon going to be a universal baseline for employment.³
Perhaps the largest operational barrier preventing enterprises from scaling AI systems successfully is a shortage of qualified internal talent. This extends beyond a scarcity of specialized machine learning engineers and data scientists to the more pervasive, everyday shortage of broad AI fluency among mid-level managers and non-technical staff who are expected to use these tools in daily operations. ⁴
Kathleen Mattie, Head of Enterprise Learning at Hartford Insurance Group puts it this way. “We are all scrambling to build talent intelligence and re-design roles on a continuous basis (forever, in my opinion). Talent intelligence gives us the ability to move employees around the organization based on skill mastery and task mastery. If we don’t know where our talent is, how can we succeed? We are all facing a data problem and the old, outdated job architecture needs to go!”
According to Deloitte’s 2026 enterprise AI tracking data, the internal skills gap is the number one barrier to successful technology integration, and while many firms have aggressively expanded their educational or training budgets, they too often focus on shallow tool tutorials rather than deep contextual problem-solving. This causes a lack of fluency that in turn creates an internal culture of hesitation, where employees either reject the technology out of fear, or use it improperly, leading to operational errors and wasted investment.
Structured upskilling strategies need to be deeply woven into career mobility paths. Training programs must emphasize practical data stewardship, prompt optimization, model evaluation, and risk identification for the average employee. For an enterprise to remain agile, its workforce must possess enough fundamental technical literacy to spot a model's hallucinations, interpret automated data readouts, and independently identify new business use cases, and the leadership of these efforts is becoming a key responsibility of human resources.
HR at Unilever
Which firms are most advanced in using human resources to manage AI integration? Unilever is one of them. Facing the challenge of managing massive volumes of global talent, Unilever’s HR department revolutionized its workforce strategy by deploying a data-driven, skills-first AI ecosystem. ⁵ Rather than viewing AI as a mechanism for headcount reduction, the firm implemented an advanced internal talent marketplace powered by predictive analytics. This system maps the existing competencies of employees and dynamically matches them with shifting operational needs and personalized, AI-driven upskilling pathways. By decoupling the concept of work from fixed job titles, Unilever’s HR team supports internal mobility and allows employees to proactively pilot AI automation tools within their respective departments. ⁶
Unilever’s approach to HR is also pioneering how it hires new employees. To manage 1.5 million annual job applications without human bottlenecking or systematic bias, the firm implemented an AI-powered system featuring neuroscience-based gamified assessments and structured video analysis as an initial filter to help assess the soft skills of applicants. This reduced the time-to-hire from four months to four weeks, while achieving a notable increase in the diversity of underrepresented hires. By strictly maintaining human oversight for all hiring decisions and continually auditing its data models for algorithmic bias, Unilever has successfully automated routine, high-volume HR transactions. This freed its human resources team to focus entirely on high-touch employee engagement and long-term strategic workforce planning.⁷
Estée Lauder did something similar by creating an internal “opportunity marketplace.” As described by Kathleen Mattie, “the system fed is by AI agents to give us every aspect of HR, from hire to retire. It includes learning and gig opportunities, skills gap analysis, workforce capability strength/weakness, and a lot more, in a personalized manner so the company wins and the employee wins. Employees, managers, business unit leaders and c-suite leaders can look at their reports through at-a-glance dashboards.”
Inevitably as AI changes the workplace and the work, it also changes the process of managing work, and pioneering HR leaders are the forefront of figuring out what it all means.
Critical Thinking
A different concern is that widespread use of generative AI will cause human critical-thinking and problem-solving skills to decline. Consequently, some organizations have introduced cognitive and skills assessments during recruitment to evaluate the capacity of potential hires for independent reasoning.
And this yet another new aspect of the evolving role of HR.
Another one, of course, is the shifting composition of the workforce as a result of the large-scale, AI-induced changes in how much of the work gets done, and the consequences for the design and operation of the entire organization.
As the days of static organizations are long gone, the HR role is morphing into an ongoing journey of organization design, re-design, and re-re-design, a continuous loop of adaption, new role definitions, new reporting models, and new skills.
SUMMARY
That’s a concise look at eight significant dimensions of change that AI is already causing, and will continue to provoke. Certainly there are more, and for your specific industry or sector the focus may shift accordingly. But the message ought to be crystal clear, that operational shifts and disruptions are going to be a permanent part of your experience going forward.
Of course all this will also have implications for what you’re actually going to do. We look at some key action items next in Part 3, which will be available soon.
REFERENCES
1 Quartz Daily Brief, July 6, 2026
2 How to Save 50 Million Euros a Year with AI
This video features the full interview where Michelin's CDIO, Yves Caseau, breaks down the company's multi-million euro AI savings strategy and specific factory floor use cases.
3 Dr. Chelsea Schein “AI Fluency Is the New Baseline. Most Hiring Teams Aren’t Ready.” April 14, 2026
4 McKinsey & Company. (2026). The State of Organizations 2026: Unlocking the AI-Enabled Organization. McKinsey Insights. https://www.mckinsey.com/capabilities/people-and-organizational-performance/our-insights/the-state-of-organizations
5 https://resources.gloat.com/resources/unilever-customer-success-story/
6 AI HR Daily Editorial. “From Four Months to Four Weeks: How Unilever and Goldman Sachs Built AI Screening Systems That Changed Enterprise Hiring.” July 10, 2026
https://aihrdaily.com/article/ai-candidate-screening-enterprise-scale-unilever-goldman-sachs-2026
7 https://airecruiterlab.com/resources/fortune-500-ai-recruitment