People and AI agents: how to plan your 2027 workforce
Also published on LinkedIn
Planning season is here. Most of the organisations we work with are building their 2027 workforce plans right now, and nearly all of them have opened last year's template: headcount by role, cost by function, a line for new hires.
The roles on that spreadsheet look the same. The work inside them doesn't.
In 2027, hybrid stops meaning office and home. It means people and AI agents doing the work side by side, formally, in the same plan. Agents take the repetitive tasks. People move up into the work that never had hands. Most 2027 plans have no row for either.
We've spent two years mapping work task by task inside live transformations. This is the guide we'd put on the table for 2027 planning.
What we're seeing in 2027 planning rooms
The plan is still a headcount spreadsheet. Leaders know AI has changed the work, but the planning tools only have rows for people. So the AI shows up as a vague efficiency target, usually a percentage cut, with no link to which work actually moved. McKinsey now says plainly that headcount-based workforce planning is no longer enough.
The junior question is the loudest and the least useful. Every planning session we sit in gets stuck on graduates. The research can't agree either: Korn Ferry has 81% of employers increasing early-career hiring, while Oliver Wyman has 43% of CEOs shifting away from junior roles. Both are counting heads. When we map the tasks, juniors aren't the only ones changing. A senior analyst checking AI output, a store manager whose rostering is now automated, a sales lead freed from reporting: every level is moving at once.
Agents are being bought, not planned. Korn Ferry found 52% of talent leaders expect to add autonomous agents to their teams. In most plans we see, those agents live in the IT budget and nowhere in the workforce plan. Nobody has written down which tasks they own, who checks their work or what happens to the hours they free.
Freed hours are disappearing. BCG reports AI scheduling saved store managers at one global retailer about four hours a week each. That's real capacity. But unless someone decides where it goes, it dissolves into email, meetings and rework. The gain is created and never banked.
The guide: Discovery, Transform, Traction
Six steps in three phases. The order matters: see the work before you change it, change it before you measure it.
Discovery: see the work
1. Map the whole organisation first. Before you plan any one team, get the full picture across the eight things that make up how work happens: people, roles, tasks, skills, processes, tools, vendors and agents. In a hybrid workforce, agents are one of the eight. They belong on the same map as everyone else. Use what you already have: role descriptions, process documents, HR and system exports, and short conversations with the people doing the work. They'll tell you what the paperwork missed. This is where the efficiencies and opportunities show up: the same work done in three teams, tasks AI is already doing unofficially, tools nobody uses, capacity sitting in the wrong place. You can't see any of that one department at a time. Output: one map of how the organisation's work gets done, and a ranked list of where to go first.
2. Then go deep, one department at a time. Start where the map shows the biggest opportunity. Break every role in that department into its tasks and the hours each takes in a typical week. Plan that department fully, learn from it, then move to the next. Each one goes faster because the map is already there. Output: a task list per role, with hours.
Transform: move the tasks, grow the roles
3. Put every task in a lane. Four lanes cover almost everything: stays human, human with AI, agent with a human checking, or stop doing it. Be honest about the fourth. Every map we build turns up work that exists only because nobody stopped it. Output: each task tagged with a lane and an owner.
4. Give agents the repetitive work, and write them into the plan. Agents take the repeating, lower-level tasks: data entry, matching, chasing, standard reports. Put each one in the plan like a team member: which tasks it handles and which person it supports. Directing and checking agents is a new skill, and a step up for whoever takes it on. Output: an agent register sitting alongside your people plan.
5. Grow every role into the freed capacity. With the routine work handled, redesign each role around the higher-level work people never had time for: analysis, judgement, customers, improvement. Plan the reskilling each person needs to get there, and redeploy capacity to where the organisation needs it most. For junior roles, keep some of the hands-on tasks that teach the craft, so they build the expertise to grow too. Accenture found one in five managers already worry juniors are missing that experience. Output: redesigned roles, each with its new task mix and a reskilling plan.
Traction: bank the hours
6. Plan capacity in hours, then headcount. Add up the hours freed and decide where each block goes: growth work the team never had time for, the backlog, quality, and time to learn. Give each decision an owner. Your strongest return is usually advancing the people you have into that work, not removing them. Headcount follows from those decisions, not the other way round. Set this as your baseline before anything rolls out, then track it quarterly from live systems, not a pulse survey. Freed time isn't ROI. Redeployed time is. Output: an hours plan with destinations, and a date for the first review.
A worked example: same team, same hours, more value
Here's the guide applied to one team: an illustrative six-person accounts payable team working 240 hours a week. We mapped its tasks, put each in a lane, then planned where the freed hours go. The figures are made up for the example, but the pattern is the one we see again and again: nobody leaves, the work moves up.
Same six people. Same 240 hours. A very different team.
Three things to notice. Exceptions go up, because the easy cases are gone and what's left needs judgement. A new task appears: someone has to own the agents and the controls around them. And the 100 freed hours don't vanish or turn into a cut. They go to work finance has wanted for years and never had hands for: better supplier terms, sharper cash forecasting, captured discounts.
The roles change with it. Processing clerks become accounts payable analysts. One person steps up as process and agent lead. The junior keeps a share of reconciliations by hand, so they still learn how the ledger works. That's a 2027 plan: every freed hour has a destination and an owner, and the team comes out more capable than it went in.
The debate worth having
The 2027 workforce debate will be framed as more people or fewer, more juniors or fewer. It's the wrong debate. The organisations that come out of 2027 ahead will be the ones that can answer a simpler question: which tasks moved, to whom, and where did the hours go?
You don't need a perfect system to start. Map the organisation, pick the department with the biggest opportunity, put its tasks in lanes and plan where every freed hour goes. Do it before the budget locks, and you'll walk into 2027 with a plan built on the work, not last year's org chart.
AI frees time. 2027 is the year you plan where it goes.
Hybrid workforces are here. Plan the work, then the people.
Sources: Korn Ferry, Class of 2030 early careers report, Oct 2026; Korn Ferry, AI in recruitment trends; Oliver Wyman Forum and NYSE, The CEO Agenda 2026; Accenture, Future of Work, Oct 2026; McKinsey, HR's dual mandate in the AI era, Jun 2026; BCG X, Frontline Ops AI.




