An evolved framework for managing capacity, costs and talent across a hybrid workforce

Workforce Planning in the Age of AI

hero image
  • 12 minute read
  • November 2026

Justin J. Hall

Partner, Workforce Solutions, PwC US

Jan Seele

Partner, Workforce Solutions, PwC US

Key takeaways:

  • Consider evolving from FTEs to effective capacity. Workforce planning may benefit from measuring the combined output of people and AI agents—not headcount alone—at the task, workflow and process level.
  • Include “bot” alongside buy, build and borrow. AI agents can serve as an additional workforce-supply lever, assessed alongside hiring, upskilling and contingent labor based on cost, quality, risk and the need for human judgment.
  • Use a more dynamic approach to hybrid workforce planning. As AI capabilities continue to develop, organizations need refreshed cost models, stronger performance data, cross-functional ownership and more frequent recalibration of the human–AI workforce mix.

INTRODUCTION There’s a gap in your workforce plan. It’s not headcount

Today, many companies are already operating a hybrid workforce of humans and AI, whether by design or by default. The question is whether their workforce planning frameworks have caught up to that reality.

While the core workforce planning process doesn't need to be reinvented, several of its key inputs, decision frameworks, and cost models must evolve. Workforce planners need to rethink how they define and measure labor supply and expand their gap-closing options. They need to reconsider how business processes are designed and update total-cost-of-workforce models to reflect a blended human-AI operating environment.

THE SHIFT Labor no longer means people – only

In a traditional workforce planning model, supply is a function of headcount, capacity, and capability. The unit of analysis is the full-time equivalent (FTE), and planning assumptions about ratios of people to workload, spans of control, and productivity benchmarks are calibrated around human workers. These ratios have been remarkably stable over time. A contact center might plan on one agent per a set number of customer interactions per hour. A finance function might staff based on transactions processed per analyst.

AI agents disrupt these ratios. Unlike human workers, they map to tasks rather than roles. A single AI agent might handle one discrete step in a workflow: extracting data from invoices, drafting initial customer responses, or flagging anomalies in a data set. The broader role remains human. The introduction of AI agents does not eliminate roles wholesale; it restructures what people within those roles spend their time doing.

This shift makes task-level decomposition essential to workforce planning. But identifying tasks AI can perform is only the starting point. Organizations must also consider how AI enables them to redesign entire workflows and processes, rather than simply automate individual activities.

For workforce planning, this has profound implications. Process transformation does not just change who does the work. It changes what work needs to be done at all.

NEW METRIC Effective capacity vs. FTE

The practical result of this shift is a new supply metric. Instead of measuring supply purely in FTE-based terms, planners need to account for blended throughput – the combination of human hours and AI agent output that together constitute the organization’s capacity to deliver work. One human working with an AI agent may produce the output that previously required two or three people, fundamentally changing the staffing ratios that have been foundational to workforce planning models.

Importantly, this does not change the demand side of the equation in a simple task-substitution model. The work still needs to get done. What changes is the translation of demand into required human headcount. When AI agents absorb portions of the workload, the number of humans needed to meet a given level of demand decreases.

The unit of planning is no longer the FTE. It’s blended throughput.

There is a talent dimension planners must not underestimate: When AI absorbs the structured, repeatable parts of a role, what remains skews toward judgment, exception-handling, relationship management, and contextual decision-making – capabilities that take years to develop and cannot be upskilled on short timelines. The implication for the “build” lever is significant: developing talent for a hybrid workforce is less about teaching people to use AI tools and more about cultivating the deeper human capabilities that become more critical as AI takes on the routine work. Career pathing, learning investments, and succession planning all need to be recalibrated accordingly.

Practically, this means organizations need to plan on two horizons. The near-term horizon addresses the current-state hybrid workforce: existing processes with AI agents absorbing selected tasks. The longer-term horizon plans for a future-state workforce shaped by transformed processes, designed from the ground up with a hybrid workforce in mind. The workforce requirements for these two states can look very different, and planning for only the first risks underinvesting in the capabilities needed for the second.

NEW COST MODEL The math has changed

Traditional total cost of workforce (TCOW) models are built around human-centric cost categories: Adding AI agents requires extending them to capture a new set of cost drivers.

AI agent costs fall into two primary categories. Implementation costs cover platform licensing, system integration, custom workflow development, and the configuration and testing required to bring agents online. Operating costs include usage-based pricing – often consumption-driven and variable – plus ongoing maintenance, monitoring, quality assurance, and the human oversight agents require to function reliably. A key component of those operating costs is token consumption. There are also hidden costs that organizations consistently underestimate. Change management, productivity dips during transition, rework during the AI agent ramp-up period, and the governance overhead of maintaining evolving AI systems all contribute to the true cost of a hybrid workforce.

The relevant comparison is the cost per unit of work delivered, adjusted for quality, reliability, speed, and scalability, rather than a simple sticker-price comparison between an AI agent and a fully -loaded human FTE. An AI agent may be dramatically cheaper per task but require meaningful human supervision that partially offsets the savings. At scale, however, an agent handling thousands of transactions without incremental labor cost may offer economics that no human staffing model can match.

Executing this analysis requires people analytics capabilities most organizations do not yet have: task-level time allocation data, real-time visibility into AI output and error rates, and productivity metrics that attribute outcomes across human and AI contributors. The best-positioned organizations will have built the infrastructure – workforce data models capturing both human and AI activity, dashboards that reflect blended throughput, and measurement frameworks that evaluate AI on comparable terms to human performance. Closing this analytics gap is itself a near-term priority.

WORKFORCE PLANNING FRAMEWORK Evolved Planning Framework: Adding a fourth "B"

When workforce supply does not meet demand, organizations have traditionally relied on three levers to close the gap: buy (hire externally), build (develop and upskill existing talent), or borrow (engage contingent workers or outsource). The emergence of AI agents introduces a fourth B – bot – and this addition interacts with the other three in ways that enrich the gap-closing decision.

Bot vs. Buy. For a given capacity gap, the first question is whether an AI agent can handle the work instead of a new hire. For highly structured, repeatable tasks with clear success criteria, the bot option may be significantly more cost-effective and faster to deploy than a traditional hire.

Bot + Build. AI agents do not only substitute for human labor; they amplify it. One person working with AI may produce what previously required two or three, making the “build” lever more powerful than it has been historically.

Bot vs. Borrow. AI agents may also serve as an alternative to contingent labor or outsourcing for certain categories of work. Where organizations have traditionally turned to staffing agencies, AI agents can offer a more controllable and scalable solution, with the added benefit of keeping institutional knowledge and data in-house.

Not all gaps should default to the bot option. Organizations need a decision framework that weighs task complexity, the degree of judgment required, regulatory and compliance constraints, risk tolerance, and organizational change readiness. The most effective approach evaluates all four levers together, recognizing that the optimal solution often involves several simultaneously: hiring a smaller number of higher-skilled humans, training them to work with AI agents, and deploying AI agents to handle the routine components of the work.

It is also worth noting that the decision to deploy AI agents creates its own talent demands. Managing agents effectively requires skills in prompt engineering, workflow design, AI oversight, and exception handling. These needs feed back into the buy and build decisions, creating a recursive loop that workforce planners must account for.

PLANNING CADENCE Plan at business speed

Traditional workforce planning is poorly suited to the pace of AI development.

Foundation models – the underlying technology powering most enterprise AI agents – are now advancing on cycles of weeks to months. Each new generation can meaningfully expand what AI agents are capable of, shift the economics of the bot option, or open entirely new categories of automatable work. When AI capabilities, costs, and adoption rates shift on a quarterly basis, a once-a-year workforce plan is already out of date before it gets approved – not because business strategy changed, but because the technology did.

The answer is to layer a monitoring-and-refresh function on top of the traditional planning cycle. When a new model release materially changes the automatable task set for a critical role, that should prompt a structured re-evaluation of staffing ratios, gap-closing decisions, and TCOW assumptions.

Organizations getting ahead are treating planning as a continuous capability that monitors, stress-tests, and recalibrates regularly – not a periodic project.

STRATEGIC CHOICE Human by design

Human effort should be intentional. That means two things: concentrating it where humans are irreplaceable – trust-based relationships, contextual judgment in ambiguous situations, inspiration and motivation, ethical navigation – and preserving it where human presence is itself the value, as a deliberate choice grounded in accountability, risk, and brand.

Work that warrants a human-by-design designation typically shares one or more of the following characteristics: accountability for the decision cannot be meaningfully assigned to an automated system; the activity is relationship-dependent in ways that affect client trust or brand perception; regulatory or legal exposure requires a named human decision-maker; or the brand risk of an AI misstep exceeds the efficiency gain.

Hybrid workforce planning also raises an important question of organizational ownership. The discipline sits at the intersection of HR, IT, operations, and finance, yet in many organizations belongs clearly to none of them. Without defined accountability, decisions about AI deployment and human workforce needs risk being made in isolation. Clarifying ownership is an essential early step.

There is also a human dimension that demands attention. How existing employees perceive and respond to AI colleagues directly affects retention, engagement, and willingness to collaborate with AI systems. Organizations that deploy AI agents without a thoughtful change management strategy risk creating anxiety and resistance that undermines adoption. When process transformation restructures roles and eliminates familiar tasks, transparent communication about the organization’s intentions becomes essential.

THE REAL RISK Inaction has a cost

Organizations that continue planning their workforce as they always have are not simply forgoing upside. They are actively accumulating structural disadvantage. Better planning leads to better deployment decisions, which generates better data, which enables more sophisticated planning.

Organizations that treat hybrid workforce planning as a separate initiative from human workforce planning will end up with fragmented strategies. The goal is integration: a single, coherent hybrid workforce plan that optimizes the blend of human and AI labor across cost, quality, risk, and strategic priorities.

Competitive advantage is not about automating the most. It lies in developing the capability to continuously optimize the human/AI blend and will go to companies that start building that capability now, before the gap becomes irreversible.

Contact us

Justin J. Hall

Justin J. Hall

Partner, Workforce Solutions, PwC US

Jan Seele

Jan Seele

Partner, Workforce Solutions, PwC US

Follow us