Accenture: The age of co-intelligence
Key takeaways
AI has progressed from a novelty to a driver of performance faster than any technology before it. Use of AI is shifting from simple augmentation, where AI supports a task, to co-intelligence, where AI can interpret intent, reason through options, coordinate steps, and execute bounded work across functions at machine speed.
Benchmarks may indicate AI surpasses people in specific domains, but only humans bring the full view, including context, values, legitimacy and accountability. Humans must stay in the lead by setting direction, defining guardrails, challenging analysis, making trade-offs and owning outcomes.
This shift raises a leadership mandate: redeploy expanded capacity into measurable value and sustained growth. Leaders need human led operating systems where people orchestrate human and AI collaboration and AI executes within clear constraints, so speed and scale increase while responsibility remains firmly human.
Accenture and The Wharton School partnered on research showing a shift to a skills first economy integrated with AI, and the report organizes value creation across four fronts: Economics, Individuals, Workforce, and Society.
Executive summary
The report argues that the next era of performance will be defined by co-intelligence: humans leading, while AI agents and robots amplify judgment, execution and autonomy. It positions this as a shift from productivity to value, and from isolated use cases to operating systems that coordinate human, digital and physical intelligence at scale.
A core message is that leadership responsibility does not shrink as autonomy rises. Instead, accountability becomes more important: humans set direction and constraints, challenge AI analysis, and own outcomes, while AI executes bounded work with machine speed.
The value agenda is framed across four fronts that leaders must integrate rather than treat as separate initiatives: economics and growth, individuals and skills, workforce design for agentic scale, and societal implications for what it means to be human at work.
Key stats and quotable claims
AI has progressed from a novelty to a driver of performance faster than any technology before it.
Use of AI has shifted from simple augmentation to co-intelligence, where AI can interpret intent, reason through options, coordinate steps, and execute bounded work across functions at machine speed.
Humans must stay in the lead by setting direction, defining guardrails, challenging analysis, making trade-offs and owning outcomes.
Leaders who master co-intelligence by integrating human, digital and physical artificial intelligence into a cohesive workforce will define how value, growth and purpose are created in the next decade.
Overall summary
This report is structured as a leadership guide to moving from tool adoption to enterprise wide reinvention. The thesis is not that autonomy replaces human work, but that it changes the operating model: work is decomposed, coordinated and executed across people, AI agents and robots, which expands capacity and compresses decision cycles.
The report emphasizes that this expanded capacity must be redeployed to create measurable value and sustained growth. That requires a human led operating system that clarifies decision rights, defines guardrails, and ensures that responsibility remains human even as execution becomes faster and more autonomous.
The deep dive below preserves substantial verbatim excerpts from the report and organizes them into the same four fronts used in the source: Economics, Individuals, Workforce, and Society.
Deep dive
The new reality: From augmentation to co-intelligence (verbatim)
AI has progressed from a novelty to a driver of performance faster than any technology before it. Use of AI has shifted from simple augmentation, where AI supports a task, to co-intelligence, where AI can interpret intent, reason through options, coordinate steps, and execute bounded work across functions at machine speed. While benchmarks indicate AI may surpass people in specific domains, only humans bring the full view, including context, values, legitimacy and accountability. That is why humans are not merely “in the loop.” Humans must stay in the lead by setting direction, defining guardrails, challenging analysis, making trade-offs and owning outcomes.
This shift raises a new leadership mandate: redeploy expanded capacity into measurable value and sustained growth. As AI compresses analysis, decision cycles and delivery, it expands both human and digital capacity, and that capacity can be redirected toward reinvention. That includes faster product iteration, new offerings, sharper customer response and smarter capital allocation. Leaders need human-led operating systems where people orchestrate human and AI collaboration and AI executes within clear constraints, so speed and scale increase while responsibility remains firmly human.
In our previous report, Humans, AI and Robots (2024), we focused on productivity: how humans and machines could work better together. This year, the focus shifts to value. Specifically, how the collaboration of human and artificial intelligence reshapes value creation across four fronts:
Economics: The next source of value is growth
Individuals: Job titles give way to skills as the new currency of work
Workforce: Architecting for agentic scale means putting people in the lead
Society: Co-intelligence will redefine what it means to be human at work
Leaders who master co-intelligence by integrating human, digital and physical artificial intelligence into a cohesive workforce will define how value, growth and purpose are created in the next decade.
Economics: The next source of value is growth
As agents move from pilots to production, the leadership question shifts from “What can AI do?” to “How much value can it create, and where does it actually come from?” Our bottom-up economic modeling shows that the agentic dividend is substantial, but it does not emerge evenly. It does, however, follow clear patterns. Leaders who understand and intentionally act on those patterns stand to capture far more value than others.
Specifically, we found that growth, not efficiency, is the dominant value lever. While productivity gains are meaningful, the primary economic upside of agentic AI comes from improved decisions, faster execution and higher-quality outcomes that translate into revenue growth. Cost reduction alone captures only a fraction of the available value.
Value is unevenly distributed across functions and tasks. Economic impact does not diffuse evenly across the enterprise. It clusters in specific functions and in a small number of task groups that recur across those functions. The implication is clear: targeted deployment, starting with priority functions and scaling the highest-value task groups across the enterprise, maximizes returns.
Productivity becomes growth only through redeployment. A significant share of productivity manifests as capacity freed rather than costs removed. Unless leaders deliberately redeploy that capacity toward higher-value work, productivity gains stall at efficiency and fail to translate into growth.
Governance needs of agentic AI depend on where tasks fall across a volume-risk spectrum. In areas where agents operate at high volumes and carry high decision risk, more rigorous governance is required. Leaders will need to assess exactly where decision risk resides before deploying agents, and design governance accordingly. This targeted approach focuses controls where they matter most, rather than applying uniform guardrails across the enterprise.
Table and chart insights, rewritten as bullets
Potential economic impact by industry: the report presents a full-maturity upper-bound annual potential of digital and physical AI agents for the average S&P 500 company, by industry, broken down into labor cost avoidance, labor cost savings and revenue uplift.
Case example of enterprise value modeling: the report applies the approach with a large enterprise with $60 billion in annual revenue and $4.5 billion in labor spend, and models approximately $6 billion in annual revenue growth alongside $1.7 billion in productivity gains.
The case example emphasizes that productivity does not translate cleanly into lower costs. Roughly two-thirds of productivity gains materialized as direct cost savings, while one-third appeared as cost avoidance, capacity freed to do different, higher-value work. Without intentional redeployment, that avoided cost does not become growth.
Individuals: Job titles give way to skills as the new currency of work
As AI reshapes work at the task level, salaries are no longer determined by job titles, but by the specific skills that drive outcomes. This shift from a role-based to skills-based labor market is the central finding of the Wharton–Accenture Skills Index (WAsX). Developed by Wharton and Accenture, WAsX is an empirical skills-mapping and economic benchmarking index that analyzes labor market dynamics at the task and skill level to measure how skills translate into economic value in an AI-enabled economy.
WAsX measures the real economics of skills by tracking four key dynamics: which skills are in oversupply or undersupply, which skills materially influence wages, how AI is reshaping skill demand over time as tasks migrate between humans and intelligent systems and how these patterns vary across industries and roles, exposing micro-economies within job families that traditional labor data often obscures.
WAsX surfaces three structural mismatches that matter directly to employers. First, AI is redistributing economic value across the skills spectrum. The demand for routine, structured cognitive skills is decreasing while the premium on skills rooted in judgment, coordination, compliance and domain-specific execution is increasing. WAsX shows that as tasks migrate between humans and intelligent systems, the market increasingly rewards skills that complement AI rather than compete with it.
Second, skills have price tags, and those prices are highly contextual. Skill value is not universal. It is governed by the micro-economics of specific roles and industries. A capability that commands a wage premium in one context may be neutral or even value-reducing in another. For employers, this means workforce strategy must move beyond generic capability frameworks toward role- and industry-specific skill economics, by identifying which skills explicitly move performance and aligning pay and development accordingly.
Third, the labor market operates with a persistent signaling gap. Workers overwhelmingly signal broad, generalist traits, while employers consistently pay for a narrower set of specialized, execution-oriented capabilities. These generalist skills are abundant and weakly differentiated, while scarce technical, analytical and operational skills remain undersupplied, creating friction in hiring and slowing execution.
For employers, these findings point to the need for structural redesign grounded in skill-level visibility. As work is increasingly decomposed into tasks and augmented by AI, leaders need systematic ways to map roles to the underlying skills that drive performance, differentiate outcomes and create economic value. Skill mapping supports precise hiring, targeted workforce development and closer alignment of compensation with the real economics of work, helping organizations invest in the capabilities that matter most.
Table and chart insights, rewritten as bullets
Skill scarcity versus wage premiums: the report illustrates how some skills are both scarce and rewarded, others remain undersupplied despite limited wage premiums, and some earn meaningful wage impact even when supply is relatively abundant. The implication emphasized is that employers need role- and industry-specific skill economics, not broad “high-value skill” assumptions.
Workforce: Architecting for agentic scale means putting people in the lead
At its core, capturing the agentic dividend is a dynamic, organizational challenge: ensuring that talent strategy keeps up with business goals and technology. The small group of companies meeting that challenge successfully today is already pulling ahead of others on several fronts. Among these companies, which we call Talent Reinventors, almost all (96%) reported having a talent strategy that is fully integrated with technology and AI through strong HR and IT collaboration.
As a result, they are reshaping work and the workforce around a shared set of goals. They are using data and AI to help them do this, but critically, they are not asking workers to layer AI onto existing processes. Nor are they demanding that people bend to new ways of working that have been established at a distance. Instead, they are reinventing what and how work gets done with humans in the lead, so that people and technology can elevate each other’s performance to meet and exceed business goals. They enable people and technology to grow, contribute and thrive together.
Six mutually reinforcing characteristics differentiate these organizations from all others. Not every Talent Reinventor is equally strong across all six. However, each characteristic is well represented across all. They are: clarity, intelligent teaming, talent mobility, co-learning, breakthrough leadership, and personalized experiences.
Already, Talent Reinventors saw revenue growth that was 1.8 percentage points higher and profit growth that was 1.4 percentage points higher than their peers in 2025. Critical to those gains, they are 7x more likely to strengthen organizational culture, 6x more likely to improve employee experience and 4x more likely to enhance workforce adaptability. They also report an 11% improvement in innovation-related skills.
Society: Co-intelligence will redefine what it means to be human at work
As individuals and organizations adapt to working alongside intelligent systems, society has an opportunity to strengthen structures that support work, learning and trust. The rise of co-intelligence is more than a technological shift. It is a chance to expand how humans create value, solve problems and contribute to collective progress.
Co-intelligence is prompting a deeper understanding of where human judgment, creativity and responsibility matter most. It invites practical questions: How can humans and AI complement one another most effectively. What forms of collaboration unlock the greatest societal benefit. How should institutions evolve to support this new partnership.
A growing body of research from Wharton faculty and other global studies suggests that as AI capabilities advance, institutions and leaders must revisit established assumptions about how work is organized and supported, while keeping accountability, legitimacy and stewardship firmly human.
The report highlights that traditional limitations of AI are being challenged across multiple dimensions, including AI as a creativity multiplier, AI as an advanced reasoner, AI as a compassionate communicator and AI as an ethical advisor.
Yet capability parity does not imply responsibility parity. While AI may generate ideas, reasoning and even ethical guidance at high levels of sophistication, it does not bear moral accountability, institutional legitimacy or long-term societal obligation. Those remain human responsibilities.
Ultimately, to unlock Human+ performance, leaders across sectors need to remove the barriers and disparities between AI haves and have-nots. The pace of change makes one-time reskilling insufficient, and new forms of collaboration between governments, academia and employers are needed to support continuous skill renewal. Adding such training could help drive significant AI investment returns and add between $4.8 and $6.6 trillion to the US economy alone by 2034.
Key takeways
The report closes with five leadership actions:
Set explicit, top-down profit and loss priorities.
Design human-led operating models for agentic work.
Reinvent the enterprise Human+ workforce.
Evaluate talent through a skills lens versus a roles lens.
Embrace opportunities as employer, educator and learner.
