Skill Partnerships in the Age of AI: How Work Is Being Reimagined Around People, Agents, and Robots

Jonathan H. Westover, PhD
7 days ago
23 min read

Listen to this article:

Abstract: The rise of artificial intelligence is fundamentally transforming the nature of work, though not in the ways most commonly feared. Rather than wholesale job displacement, what's emerging is a new partnership model where people collaborate with AI agents and robots, each contributing distinct capabilities. Drawing on analysis of approximately 800 US occupations and 2,000 detailed work activities, this research reveals that while current technologies could theoretically automate 57% of US work hours, most human skills will remain essential—though applied differently. More than 70% of today's skills span both automatable and non-automatable work, meaning they will evolve rather than disappear. The economic stakes are substantial: by 2030, reimagining workflows around people-agent-robot collaboration could unlock approximately $2.9 trillion in US economic value. However, capturing this potential depends less on technological breakthroughs than on deliberate organizational choices—redesigning complete workflows rather than automating isolated tasks, investing in workforce capabilities alongside technology, and building new performance frameworks that account for human-machine collaboration. This article examines how AI is reshaping skill demands across occupations, introduces a Skill Change Index to measure exposure to automation, and presents evidence from early adopters who are successfully reimagining work. The findings suggest that leadership commitment, cultural transformation, and sustained investment in human capabilities will ultimately determine whether AI expands opportunity or concentrates it.

The conversation about artificial intelligence and work has reached an inflection point. As AI systems demonstrate increasingly sophisticated capabilities—from generating clinical reports to orchestrating sales workflows—anxiety about job displacement has intensified. Yet this framing misses the more fundamental transformation underway: AI is not simply replacing human workers but reshaping what work means and how value is created.

Consider the experience of a regional bank that recently deployed AI agents to modernize its legacy software systems. Previously, such projects required months of manual documentation, code refactoring, and testing by large engineering teams. With AI agents handling assessment, code migration, and automated testing, human developers shifted from executing repetitive tasks to orchestrating the process—setting architectural requirements, validating outputs, and ensuring functional accuracy. The result: up to 70% code accuracy and an estimated 50% reduction in required human hours, with developers freed to focus on strategic design rather than line-by-line implementation (Yee et al., 2025).

This pattern is appearing across industries. A global pharmaceutical company cut clinical study report drafting time by nearly 60% while reducing errors by roughly 50% using AI companions that synthesize study data and generate comprehensive drafts in minutes. A major utility company now resolves approximately 40% of customer service calls through agentic conversational AI, cutting average cost per call by about 50% while improving customer satisfaction scores. In each case, the technology didn't eliminate roles—it reconfigured them around a new division of labor between human judgment and machine capability (Yee et al., 2025).

These examples point toward a future where work is organized as a partnership between people, AI agents that handle nonphysical cognitive tasks, and robots that perform physical activities. This article examines how this partnership is taking shape, what it means for skills and occupations, and how organizations and institutions can navigate the transition effectively.

The Workforce Landscape: A Partnership Emerges

Redefining Automation Potential

For much of the past century, machines followed rules. Industrial robots executed predetermined physical routines; software automated predictable clerical tasks. Both operated within narrow parameters, doing what they were programmed to do and little more. The rise of AI—particularly generative AI and large language models—has begun to change this fundamental limitation. These systems can learn from vast datasets, simulate reasoning, respond to natural language, and adapt to varied contexts rather than simply following preset instructions (Yee et al., 2025).

This leap in capability has dramatically expanded what can be automated. McKinsey Global Institute's updated analysis estimates that current technologies could, in theory, automate activities accounting for about 57% of US work hours today. This figure reflects the technical frontier—what is possible given demonstrated technologies, including those still in laboratory settings—not a prediction of job losses. It is an assessment of potential, not inevitability (Yee et al., 2025).

The distinction matters. Technical potential measures the ceiling of what machines can do; actual adoption depends on implementation timelines, relative costs of technology versus labor, organizational readiness, regulatory frameworks, and cultural acceptance. History shows that widespread diffusion typically takes time. Electricity took more than 30 years to spread through manufacturing; industrial robotics followed a similar multi-decade trajectory. As recently as 2023, only about one in five companies ran most applications in the cloud despite the technology being widely available since the mid-2000s (Yee et al., 2025).

The Division of Work: People, Agents, and Robots

Understanding how AI will reshape work requires distinguishing between physical and nonphysical activities. Agents—AI systems that handle cognitive work—and robots—machines that perform physical tasks—represent different automation frontiers with different timelines and constraints.

Nonphysical work accounts for roughly two-thirds of US work hours. Within this domain, approximately one-third draws on social and emotional skills that remain largely beyond AI's reach—reading a student's expression mid-lesson, sensing when a client is losing interest, building trust through authentic human connection. The remaining nonphysical work involves tasks like reasoning, analysis, document creation, and information processing that are increasingly automatable. These more automatable activities represent about 40% of total US wages and span occupations from accountants and lawyers to teachers and healthcare administrators (Yee et al., 2025).

Physical work presents a narrower near-term automation opportunity. Activities requiring physical capabilities account for about 35% of current US work hours, but most still demand fine motor skills, dexterity, and situational awareness that technology cannot yet replicate reliably. Advances continue—new generations of general-purpose robots powered by AI can increasingly operate in unstructured environments, follow verbal instructions, and execute task variations they weren't explicitly trained on. Yet major barriers remain, including dexterity limitations, safety concerns when operating near humans, power constraints (most humanoids operate untethered for only two to four hours per charge), and cost (per-unit costs of 150,000–150,000–150,000–500,000 in the US would need to fall to roughly 20,000–20,000–20,000–50,000 for large-scale adoption) (Yee et al., 2025).

Even so, the effects on certain workers could be significant. Physical tasks make up more than half of working hours for about 40% of the US workforce, including drivers, construction workers, cooks, and healthcare aides. As robotics technology matures, these occupations will experience substantial change, particularly in production and food preparation roles that include many lower-wage positions. Robots may also increasingly perform work that is hazardous or otherwise unfeasible for people—underwater inspections, search-and-rescue operations, dangerous facility checks (Yee et al., 2025).

Seven Occupation Archetypes

To map how automation potential varies across the economy, the research analyzed approximately 800 US occupations based on the share of work hours that could be performed by people, agents, or robots. This analysis yielded seven distinct archetypes that illustrate the range of human-machine collaboration models likely to emerge (Yee et al., 2025).

People-centric occupations (34% of current US workers, $71,000 average annual pay) remain largely human-centered. Registered nurses, psychologists, and firefighters fall into this category—roles where physical tasks that current technologies cannot replicate account for about half of work hours, supplemented by substantial social and emotional work. These occupations will incorporate AI tools at the margins but fundamentally remain dependent on human capabilities (Yee et al., 2025).

Agent-centric occupations (30% of workers, $70,000 average pay) involve large shares of cognitive work that AI can increasingly handle—drafting documents, analyzing data, conducting research. Accountants, software developers, and lawyers exemplify this archetype. While automation potential is high, people remain essential to set direction, validate outputs, handle exceptions, and exercise professional judgment. The work becomes more supervisory and strategic as agents take on routine cognitive tasks (Yee et al., 2025).

Robot-centric occupations (8% of workers, $42,000 average pay) involve physically demanding, often hazardous work that could theoretically be largely automated once robotic capabilities advance sufficiently and costs decline. Stockers and order fillers, welders, and cooks fall into this category. In practice, cost and capability constraints mean people will remain central for years, though their roles will shift toward oversight, quality control, and handling tasks requiring human dexterity (Yee et al., 2025).

Between these poles lie hybrid archetypes that combine people, agents, and robots in varying configurations:

People-agent roles (21% of workers, $74,000 average pay): Sales representatives, secondary school teachers, and HR specialists whose work could be enhanced by digital and AI tools while remaining fundamentally human-centered
People-robot roles (<1% of workers, $54,000 average pay): Insulation workers, drywall installers, and similar construction/maintenance roles where machines add strength and precision to human efforts
People-agent-robot roles (5% of workers, $60,000 average pay): Receptionists, medical assistants, and correctional officers whose work combines physical presence, cognitive tasks, and interpersonal interaction
Agent-robot roles (2% of workers, $49,000 average pay): Machine setters, bakers, and library assistants in production settings where software intelligence directs physical systems

These archetypes reflect technical automation potential under current capabilities, not predictions of actual future work models. As technologies evolve and organizations adapt workflows, the distribution of roles and the nature of collaboration within each archetype will continue to shift. What remains constant is that people, agents, and robots each bring distinct strengths—and that value creation increasingly depends on combining them effectively (Yee et al., 2025).

Table 1: Archetypes of Human-Machine Collaboration in the AI Era

Occupation Archetype	Share of US Workforce (%)	Average Annual Pay	Primary Work Components	Automation Potential (Inferred)	Example Roles	Strategic Evolution of Roles
People-centric	34%	$71,000	Physical tasks (50%) and high social-emotional work	Low; limited by need for human connection and non-replicable physical skills	Registered nurses, psychologists, firefighters	Incorporating AI tools at the margins while remaining dependent on human capabilities
Agent-centric	30%	$70,000	High cognitive work (drafting, analysis, research)	High; core cognitive tasks like data processing and drafting are increasingly automatable	Accountants, software developers, lawyers	Shifting from execution of routine cognitive tasks to orchestration and professional judgment
People-agent	21%	$74,000	Mix of cognitive tasks and high interpersonal interaction	Moderate; AI agents augment research while human social skills remain vital	Sales representatives, secondary school teachers, HR specialists	Work is enhanced by AI tools while remaining fundamentally human-centered
Robot-centric	8%	$42,000	Physically demanding and often hazardous work	High; technically feasible as robotic costs decline and dexterity improves	Stockers and order fillers, welders, cooks	Shifting toward oversight, quality control, and tasks requiring high human dexterity
People-agent-robot	5%	$60,000	Combines physical presence, cognitive tasks, and interpersonal interaction	Moderate-High; requires coordination across multiple automation domains	Receptionists, medical assistants, correctional officers	Orchestrating various systems while providing necessary physical or human touchpoints
Agent-robot	2%	$49,000	Software intelligence directing physical systems in production	High; relies on integration of cognitive agents and physical machinery	Machine setters, bakers, library assistants	Shifting from manual machine operation to managing software-directed physical systems
People-robot	<1%	$54,000	Construction and maintenance requiring strength and precision	Moderate; machines handle heavy lifting while humans handle navigation and installation	Insulation workers, drywall installers	Machines add strength and precision to human efforts

The Evolution of Skills: Adapting Rather Than Disappearing

The Skill Requirements Landscape

Perhaps the most consequential—and least understood—dimension of AI's impact concerns skills. While much public attention focuses on whether jobs will disappear, the more pressing question is how the fundamental building blocks of work will change. Skills represent the knowledge, competencies, and attributes people deploy to perform work activities. As AI takes on more tasks, skill requirements are becoming both more specific and more fluid (Yee et al., 2025).

Analysis of job postings reveals that the number of distinct skills associated with each occupation has risen on average to 64 from 54 a decade ago, reflecting greater specificity in how employers describe roles. This trend toward specialization correlates with wages: higher-paying fields tend to require more skills and greater specialization. Data scientists and economists, for instance, show more than 90 unique skills in job postings, compared with fewer than ten for motor-vehicle operators (Yee et al., 2025).

The skill mix also varies systematically by occupation type. Higher-wage jobs emphasize management, information processing, and digital capabilities. Lower-wage roles focus on hands-on work, operating equipment, and providing care and assistance. Yet even within a single field like software development, required skills diverge sharply by specialization: Python developers, AI engineers, and C++ developers share fewer than half their required skills, illustrating how technology drives fragmentation (Yee et al., 2025).

This increasing specificity carries a paradox: as skills multiply and specialize, the importance of transferable skills grows. Eight high-prevalence skills—communication, management, operations, problem-solving, leadership, detail orientation, customer relations, and writing—remain essential across industries and wage levels. These skills form the connective tissue of the labor market, enabling workers to move between roles and adapt to changing requirements. Their application will evolve as people work more closely with AI, but their relevance endures (Yee et al., 2025).

The Surge in AI-Related Skill Demand

As AI technology matures, related skill requirements are spreading beyond development roles at unprecedented speed. Demand for AI fluency—the ability to use and manage AI tools—jumped nearly sevenfold in the two years through mid-2025, making it the fastest-growing skill category in US job postings. This surge dwarfs the growth rate of any other skill cluster and likely marks the beginning of much larger changes ahead (Yee et al., 2025).

AI fluency encompasses two primary capabilities: using AI, which involves leveraging AI tools and applications in workflows for everyday tasks, and managing AI, which involves managing hybrid human-agent-robot teams and orchestrating workforce design and strategy. Together, these skills now appear as requirements in occupations employing about seven million US workers—a fraction of what may ultimately be needed (Yee et al., 2025).

Demand for technical AI skills—building and deploying AI systems—has also grown, though at a slower pace than fluency requirements. Technical AI skills include two domains: developing AI, which requires engineering systems with deep technical expertise, and governing AI, which ensures responsible, ethical, and compliant AI deployment. The combination of fluency and technical skills has expanded to occupations employing about 7.5 million workers (Yee et al., 2025).

Yet AI skill demand remains concentrated. Three-quarters of all AI skill requirements in the United States appear in just three occupational groups: computing and mathematics; management; and business and finance. Ten additional groups show emerging demand, including architecture and engineering, installation and maintenance, and education. Nine other occupational groups—construction, transportation, food service, and others accounting for roughly 40% of the workforce and falling below median income—show virtually no AI skill requirements in current job postings (Yee et al., 2025).

This concentration suggests both opportunity and risk. Industries and roles already adopting AI aggressively are building capabilities and experience that may compound over time. Workers in occupations not yet touched by AI skill requirements may face steeper learning curves and greater disruption when requirements eventually arrive. Closing this gap will require coordinated investment in training and workforce development before demand surges (Yee et al., 2025).

Skills That Are Rising, Evolving, or Declining

Beyond AI-specific capabilities, broader patterns are emerging in skill demand. Employers increasingly seek AI-adjacent skills such as quality assurance, process optimization, and teaching—capabilities needed to redesign work with AI, supervise and verify AI systems, or train people to use them effectively. These complementary skills are growing steadily as organizations move beyond simply deploying tools to fundamentally restructuring workflows (Yee et al., 2025).

Certain physical skills are also experiencing renewed demand, particularly nursing and electrical work. This uptick may reflect sectors where physical presence and dexterity remain essential—healthcare, skilled trades, infrastructure—and where labor shortages persist despite automation's advance. These roles often command middle-income wages and offer clear career pathways, making them attractive destinations for workers displaced from more automatable occupations (Yee et al., 2025).

At the same time, job posting mentions are declining for skills where AI already performs well or significantly augments human work. Research, writing, and simple mathematics show falling mention rates, though these skills remain essential for much of the workforce. The decline likely reflects employer expectations that AI tools will handle routine applications of these skills, reducing the need to specify them explicitly while raising the bar for demonstrating mastery that adds value beyond what AI provides (Yee et al., 2025).

This pattern—rising demand for skills that complement or guide AI, stable demand for skills requiring human judgment or physical presence, and falling demand for skills AI replicates—points toward a reshuffling rather than a wholesale elimination of skill requirements. Workers will need to apply familiar skills in new contexts, develop fluency with AI tools, and increasingly emphasize capabilities that machines cannot match.

The Skill Change Index: Measuring Exposure

To quantify how different skills may be affected by automation, the research introduces the Skill Change Index (SCI)—a time-weighted measure of each skill's potential exposure based on the automation adoption rates projected for work activities where that skill is applied (Yee et al., 2025).

The SCI reveals three broad categories:

People-led skills (11% of mapped skills) are required predominantly for work that cannot be automated with current technologies. These are often rooted in social and emotional intelligence—conflict resolution, relationship management, design thinking, coaching. They depend on empathy, creativity, and contextual understanding that remain beyond machine capability. Skills in this category face the lowest exposure to change (Yee et al., 2025).

AI-led skills (17% of mapped skills) are associated primarily with automatable activities. Data entry, financial processing, equipment control, and certain programming languages fall into this category. These are often specialized skills where AI already performs reliably or is advancing rapidly. People working in these areas will likely step back from hands-on execution to focus on designing systems, validating results, and handling exceptions (Yee et al., 2025).

Shared skills (72% of mapped skills) are required for both automatable and non-automatable work. This large middle ground represents the emerging partnership: machines handle routine tasks while people frame problems, interpret results, and make decisions. The work blends collaboration and oversight, with humans providing judgment and contextual understanding that machines lack. Most of the eight high-prevalence skills identified earlier—communication, problem-solving, management, detail orientation—fall into this category (Yee et al., 2025).

Within the shared category, exposure varies significantly. Under the midpoint adoption scenario, approximately one-quarter to one-third of work hours tied to the 100 most in-demand skills could be automated by 2030. For example, about 28% of work associated with quality assurance could be performed by machines in this timeframe. In a faster adoption scenario, exposure rises sharply—the most affected skills among the top 100 could reach 60% automation, while quality assurance could hit 50% (Yee et al., 2025).

Examining the broader universe of approximately 6,800 mapped skills, digital and information-processing skills rank highest on the SCI, reflecting AI's growing proficiency in data handling and analysis. Assisting and caring skills—basic first aid, patient care, peer support—rank lowest, consistent with their concentration in people-led work. Between these extremes, most skills show moderate exposure, indicating they will evolve in application rather than simply rise or fall in demand (Yee et al., 2025).

The SCI framework suggests three distinct paths:

Highly exposed skills (top quartile) are likely to decline in demand as AI handles them reliably—think accounting processes or programming in specific languages that AI already performs well
Moderately exposed skills (middle quartiles) are likely to evolve, changing in nature and application rather than disappearing—writing, research, and quality assurance fall here, remaining relevant but applied differently as AI becomes a collaborative partner
Low-exposure skills (bottom quartile) are likely to endure with minimal change, grounded in human connection and care—leadership, interpersonal communication, and healthcare skills exemplify this category

As AI capabilities advance and adoption accelerates, some partially automatable skills may become more exposed while entirely new skills emerge. The SCI offers a snapshot of current exposure, not a permanent classification. What it reveals most clearly is that nearly all occupations will experience skill shifts by 2030, and most human capabilities will remain relevant even as their application transforms (Yee et al., 2025).

Reimagining Workflows: Where Economic Value Resides

From Task Automation to Workflow Redesign

The difference between disappointing and transformative AI outcomes often hinges on scope of ambition. Organizations that apply AI to individual tasks within existing processes—adding a chatbot for ad hoc employee questions, for instance—typically see modest benefits. Those that redesign entire workflows around new divisions of labor between people, agents, and robots unlock substantially larger gains (Yee et al., 2025).

This distinction explains a puzzling gap: nearly 90% of companies report investing in AI, yet fewer than 40% report measurable gains. The shortfall likely reflects two factors. First, many projects remain in pilot or trial phases, not yet scaled across operations. Second, many applications target discrete tasks rather than reimagining complete processes. In banking, this is the difference between offering a chatbot for ad hoc use versus deploying custom agents alongside people in a reimagined loan approval, processing, and management workflow that delivers faster decisions and better customer service (Yee et al., 2025).

Workflows—multistep processes involving collaboration, information exchange, and decision-making—form the backbone of how organizations operate. Most were designed for a pre-AI world. Retrofitting them with AI at the margins cannot deliver the productivity transformation now possible. Capturing that potential requires rethinking workflows from first principles: What outcomes do we seek? Which activities require human judgment? Where can agents and robots operate autonomously? How should handoffs and oversight work? What new roles and skills do people need? (Yee et al., 2025).

The $2.9 Trillion Opportunity

Analysis of approximately 190 business processes across the US economy identifies where the greatest opportunities lie. In the midpoint adoption scenario for 2030, AI-powered agents and robots could generate roughly $2.9 trillion in annual US economic value—calculated as the value of work hours automated multiplied by associated wages. About 60% of potential productivity gains concentrate in sector-specific domains at the core of each industry: supply chain management in manufacturing, clinical diagnosis and patient care in healthcare, regulatory compliance and risk management in finance. The remaining 40% comes from cross-cutting functions—IT, finance, administrative services—that support every sector (Yee et al., 2025).

This economic value reflects resources that automation could release and redirect to other productive uses. Whether these translate into GDP growth, productivity gains, or employment effects depends on how organizations and the broader economy redeploy freed capacity—a question beyond this analysis's scope but central to determining AI's ultimate impact.

The value varies significantly by sector. Finance and insurance show particularly high potential given the industry's cognitive intensity and data richness. Manufacturing, healthcare, and professional services also show substantial opportunity, though the mix of agent versus robot contribution differs. Construction, transportation, and food services show more modest potential given current technology limitations around physical work, though this will shift as robotics capabilities advance (Yee et al., 2025).

Within sectors, certain workflows show disproportionate promise. Customer service and technical support, for instance, represents a large-scale opportunity across industries as conversational AI matures. Software application development offers substantial value as coding agents accelerate both new development and legacy system modernization. Medical research and documentation support could transform healthcare timelines. Lead generation and qualification in sales, inventory planning and optimization in logistics, and financial planning and analysis in corporate finance all represent high-impact workflows ripe for redesign (Yee et al., 2025).

Cases from the Field

Four implementation examples illustrate how workflow redesign is unfolding in practice, each demonstrating the shift from task automation to systemic transformation.

Sales: Expanding Reach Through Agent Orchestration

A global technology company sought to deepen customer relationships across thousands of smaller accounts that previously received inconsistent attention due to limited sales capacity. Rather than simply providing sales teams with better tools, the company introduced a coordinated system of AI agents that automate early-stage sales activities (Yee et al., 2025).

A prioritization agent scores and ranks accounts using public and proprietary data. An outreach agent contacts customers with tailored messaging. A customer response agent manages follow-ups, categorizing leads as interested, not interested, or uncertain. A scheduling agent sets up calls and reminders for high-potential leads. When a case requires human judgment, a handoff agent transfers the file to a business development specialist with full context and history (Yee et al., 2025).

This process expanded outreach dramatically while improving conversion rates, delivering a projected annual revenue increase of 7% to 12% from new sales, cross-selling, and retention. Time saved across sales roles ranged from 30% to 50%, allowing specialists to redirect effort from routine tasks toward strategic engagement—drafting proposals, negotiating partnerships, building relationships. Looking forward, the company envisions adding a coaching agent to provide real-time feedback and an admin agent to handle routine administrative tasks, further elevating human roles (Yee et al., 2025).

Customer Service: Resolving Issues at Scale

A large utility company handles more than seven million support calls annually, even with multiple self-service options available. Its interactive voice response system previously resolved only about 10% of inquiries, leaving the rest to human representatives. To improve efficiency and customer experience, the company deployed agentic conversational AI across its entire customer base (Yee et al., 2025).

The system includes multiple specialized agents: an inbound call agent authenticates customers, an intent identification agent determines the call's purpose, a call scheduling agent manages appointments, and a self-service agent integrates with back-end systems to resolve issues. Together, these now handle roughly 40% of all calls, resolving more than 80% without human involvement. When escalation is needed, customers transfer to representatives with verified account details and conversation history, ensuring seamless handoffs (Yee et al., 2025).

The results: average cost per call fell by approximately 50%, customer satisfaction scores rose six percentage points driven by shorter wait times and faster resolution, and human representatives now manage more complex, emotionally sensitive, and high-value issues. Future applications could extend this further—a customer issue identification agent could proactively contact customers when service interruptions are detected, while a coaching agent could provide real-time guidance to representatives during live calls (Yee et al., 2025).

Medical Writing: Accelerating Drug Development

A global biopharmaceutical company sought to improve the process for drafting clinical study reports—lengthy documents that compile safety and efficacy data for new drugs. Traditionally, medical writers manually assembled study data, drafted reports, and coordinated multiple review cycles. Limited capacity and long turnaround times constrained the ability to meet growing regulatory submission demands (Yee et al., 2025).

The company developed an AI platform that reconfigures the reporting workflow. A data mapping agent synthesizes structured and unstructured study data. A report-drafting agent generates comprehensive first drafts in minutes, applying company style and compliance templates. A validation agent checks data accuracy and regulatory compliance. A reviewing agent regenerates sections based on feedback, enabling rapid iteration. Throughout, medical writers collaborate with the AI—validating outputs, adding clinical judgment, ensuring cohesive narrative, and maintaining scientific rigor (Yee et al., 2025).

Early data indicate substantial efficiency gains: touch time for first human-reviewed drafts dropped nearly 60%, errors declined roughly 50%, and go-to-market timelines accelerated by weeks when combined with other process improvements. The company reports that scaling these efforts requires a combination of technology advancement and workforce upskilling—resilient data engineering, prompt engineering capabilities among writers, and bold organizational leadership willing to transform established practices (Yee et al., 2025).

IT Modernization: Shifting Developers to Orchestration

A regional lender used AI agents to modernize its banking application for small and medium enterprises, aiming to update legacy programming languages and speed internal development. The project would previously have required months of work, large budgets, and extensive engineering capacity for manual documentation, code refactoring, and testing of millions of lines of code (Yee et al., 2025).

The bank launched a pilot using AI agents for multiple modernization tasks. An assessment agent scans legacy code bases identifying dependencies. A functionality agent generates target-state architecture. A coding agent migrates code to new frameworks and performs automated tests. Human developers collaborate with 15 to 20 agents each, verifying and refining outputs to ensure architectural integrity, compliance, and functional accuracy. The modernization also shifted applications from desktop to mobile, on-premises to cloud, and monolithic to microservice architectures (Yee et al., 2025).

As agents took on most repetitive execution, human work shifted toward planning, orchestration, and testing. Early results show up to 70% code accuracy. Following the pilot, the bank estimates extending agent use to the entire effort could reduce required human hours by up to 50%. Additional agents under consideration include a modernization planning agent to coordinate the overall process, quality assurance agents to systematically test outputs, and testing agents to automate validation (Yee et al., 2025).

The Transformation of Management

These cases reveal a consistent pattern: as AI takes on analytical and decision-support tasks, the nature of managerial work shifts from supervising people to orchestrating systems in which people, agents, and robots collaborate. This change allows managers to redirect time to higher-value work involving coaching, influencing, and mentorship, while also demanding greater technical fluency (Yee et al., 2025).

A sales manager might spend less time on pipeline tracking (now automated) and more time coaching teams to interpret AI-driven insights, refine customer engagement strategies, and strengthen relationships. A customer service manager might oversee a hybrid workforce of people and AI agents, training both AI systems and staff to improve service quality. An IT manager might coordinate agent-led development sprints, validate architectural decisions, and ensure systems meet security and compliance requirements (Yee et al., 2025).

This evolution affects multiple management skills differently. Prioritization becomes more dynamic, with AI agents sequencing tasks based on real-time data while humans balance stakeholder needs and apply strategic judgment. Decision-making expands to include evaluating AI-simulated scenarios and choosing among options machines generate. Planning scales through AI's ability to reallocate resources and orchestrate workflows, with humans focusing on design and exception handling. Accountability shifts toward interpreting AI audit trails and ensuring model integrity rather than directly observing work completion (Yee et al., 2025).

The implication: organizations must prepare managers not just to use AI tools but to lead in an environment where value creation depends on effectively combining human and machine contributions. This requires new competencies, new performance metrics, and new mental models of what management means.

Leadership Imperatives: Navigating the Transition

Questions for Business Leaders

Embedding AI successfully depends on recognizing the enduring importance of people—not as a platitude but as a practical and ethical necessity. As technology takes on more tasks, the judgment and oversight people provide becomes more vital for keeping organizations on course. Several key questions help frame the choices leaders face (Yee et al., 2025).

Are you reimagining your business for future value? Early AI efforts often aim to improve existing workflows rather than rethink them. Larger gains come from redesigning processes entirely, which means looking several years ahead and working backward to identify which roles, skills, and structures need to change. Leaders must choose where to invest in major redesigns now versus refining current models for nearer-term gains (Yee et al., 2025).

Are you leading AI as a core business transformation? AI will touch nearly every function. Leaders can approach it as a technology project or as a broader business transformation. Delegating to IT may speed implementation, but lasting change and real strategic advantage depend on visible commitment from senior leadership and sustained attention to how AI affects people and operations across the organization (Yee et al., 2025).

Are you building a culture of experimentation and learning? Implementing AI involves uncertainty. Organizations that test and adapt quickly tend to learn fastest. This depends on culture that supports curiosity, risk-taking, learning from setbacks, and collaboration. Changing culture is difficult but essential for transformation on the scale AI requires (Yee et al., 2025).

Are you building trust and ensuring safety? AI changes how businesses maintain accountability. The focus shifts from checking individual outputs to setting policies, validating AI logic, handling exceptions, and determining when human involvement is most needed. The challenge is maintaining sufficient oversight to manage risk and ensure safety without limiting efficiency and innovation (Yee et al., 2025).

Are you equipping managers to lead teams of people, agents, and robots? AI redefines management. Routine supervision may be automated, freeing managers for coaching, influencing, and orchestrating hybrid teams. They also play key roles in testing for bias, validating performance, and upholding integrity. As direct control diminishes, staying accountable for outcomes becomes more challenging, requiring new performance metrics and feedback systems (Yee et al., 2025).

Are you preparing workers for new skills and roles? Companies must decide how to use capacity freed by AI—whether to reinvest in developing people and higher-value work or focus on efficiency and cost reduction. Most will do both. Managing this shift means identifying which roles can evolve and giving employees clear, skill-based pathways to grow into them. AI fluency must extend across all organizational levels, and companies can use digital tools, hands-on projects, coaching, and partnerships with other institutions to build these capabilities (Yee et al., 2025).

Questions for Institutions

Periods of economic disruption often force societies to strengthen systems that help people adapt. The rise of AI may call for similar renewal, with public, private, and civic institutions leading by example in retraining people and expanding opportunity (Yee et al., 2025).

How can education and training keep pace? Foundations of AI fluency—critical thinking, questioning results, challenging assumptions, recognizing bias—should be developed from primary school onward. Curricula could be redesigned to combine technical knowledge with transferable skills like adaptability, analytical thinking, and collaboration. Universities might integrate AI across disciplines while vocational and community colleges expand training in skilled trades. AI itself could support more personalized and continuous learning. As demand for reskilling grows, education systems and employers may need to work more closely through shared programs, flexible models, earn-as-you-learn apprenticeships, and faster credentialing (Yee et al., 2025).

What systems are needed to ensure transferable skills lead to opportunities? As AI transforms work, many people will need to move into entirely new occupations. Transferable skills will be essential, but they matter only if the labor market can recognize and reward them. Clear skill definitions, trusted demonstration methods (testing or verified credentials), and better matching platforms could make this possible. Building links between employers, schools, and credentialing institutions could expand access to work and opportunity (Yee et al., 2025).

How can local economies and communities respond? AI's impact will vary widely across industries and regions. Understanding differences through data is the first step toward effective action. With clear pictures of where change is happening, industry groups, educators, workforce agencies, and unions can collaborate on training and job-transition strategies that fit local needs (Yee et al., 2025).

Conclusion

The partnership between people, agents, and robots is already taking shape as businesses embed technologies in workflows and skill profiles shift across industries. What distinguishes this technological moment from previous ones is not the pace of change—though rapid—but the breadth. AI reaches into reasoning, communication, and judgment, touching work once considered beyond automation's frontier.

Yet the evidence suggests that most human skills will remain relevant, applied differently rather than rendered obsolete. More than 70% of today's skills span both automatable and non-automatable work, positioning them to evolve as people collaborate more extensively with intelligent machines. The eight high-prevalence skills that cut across occupations and industries—communication, management, problem-solving, customer relations, leadership, detail orientation, operations, and writing—will continue to matter, though their application will shift as AI handles routine dimensions of each.

The economic opportunities are substantial. Capturing the estimated $2.9 trillion in US economic value by 2030 depends less on technological breakthroughs—current capabilities already exceed widespread adoption—than on organizational willingness to redesign workflows, invest in workforce capabilities alongside technology, and build performance frameworks that account for human-machine collaboration. Early movers demonstrate this is achievable: sales organizations expanding reach through agent orchestration, customer service operations resolving issues at scale, pharmaceutical companies accelerating drug development, banks modernizing legacy systems. In each case, the technology did not eliminate roles but reconfigured them around new divisions of labor.

The transition will not be smooth or equitable by default. AI skill demand remains concentrated in higher-wage occupations and knowledge-intensive sectors. Workers in construction, transportation, food service, and other physical occupations—representing roughly 40% of the US workforce below median income—show virtually no AI skill requirements in current job postings. When requirements eventually arrive, these workers may face steeper learning curves absent proactive investment in training and capability building. Narrowing this gap before disruption intensifies represents both moral imperative and economic necessity.

Leadership—in businesses, educational institutions, workforce agencies, and government—will largely determine whether AI expands opportunity or concentrates it. The most effective leaders will engage directly with AI rather than delegating, invest in human skills that complement rather than compete with technology, and balance productivity gains with responsibility for worker welfare and community resilience. They will recognize that technology adoption is not an end in itself but a means toward expanding human potential and creating broadly shared prosperity.

The partnership model emerging between people, agents, and robots offers a more optimistic vision than wholesale job displacement. It positions humans at the center of work—not despite AI but because of it. Machines still depend on human guidance, interpretation, quality control, and accountability. They lack contextual understanding, ethical reasoning, and the capacity for authentic connection that remain distinctly human. The question is not whether people will remain essential but how organizations and societies will prepare them for roles that increasingly involve orchestrating intelligent systems rather than executing individual tasks.

Today's technologies offer vast opportunities to increase productivity and enhance human capabilities. How work evolves depends on choices made now—about where to invest, how to redesign workflows, which skills to prioritize, and whether benefits are widely shared. The future is not predetermined. It will be shaped by the deliberate actions of leaders willing to reimagine work as a partnership where people, agents, and robots each contribute their strengths toward common purpose.

Research Infographic

AI Era Value Creation Strategy Slide Deck

References

Yee, L., Madgavkar, A., Smit, S., Krivkovich, A., Chui, M., Ramirez, M. J., & Castresana, D. (2025). Agents, robots, and us: Skill partnerships in the age of AI. McKinsey Global Institute.

Jonathan H. Westover, PhD is Chief Research Officer (Nexus Institute for Work and AI); Associate Dean and Director of HR Academic Programs (WGU); Professor, Organizational Leadership (UVU); OD/HR/Leadership Consultant (Human Capital Innovations). Read Jonathan Westover's executive profile here.

Suggested Citation: Westover, J. H. (2026). How Emerging Technologies Can Foster Human Connections at Work. Human Capital Leadership Review, 32(4). doi.org/10.70175/hclreview.2020.32.4.4