Roles & the shape of the workforce

Which roles change, which appear, and what the org chart looks like after.

33documents on this topic
15organizations represented
4issues named
10sourced citations
0sourced statistics

The state of it

One of 4 topics within Operating model redesign.

The largest empirical base in this theme is a study of over one billion job advertisements, and its central finding is a fork: AI is splitting work into two tracks. In 52% of advertised jobs the technology takes on the complex tasks - democratising the role, and flattening its prospects. In 22% it takes the basic work and leaves the expert tasks - professionalising the role. The professionalised track is growing twice as fast and its advertised pay has grown 42% faster since 2021. Which track a role lands on is partly a design choice, which is what makes this an operating-model topic rather than a labor-market observation.

The redesign is sharpest at the bottom rung. Early-career postings have flatlined only in the most AI-exposed quartile of occupations, and the entry jobs that remain are being rebuilt upward: the most exposed junior roles are seven times more likely than the least exposed to demand traditionally senior skills, and the redesigned versions are growing 35% while the unredesigned ones decline. The apprenticeship model - years of basic tasks as training - is quietly disappearing, and nothing structured has replaced it.

Above the entry level, a wave of new roles is arriving without a place on the chart: 78% of leaders say they are considering AI-specific hires, and the planning unit itself is shifting - from headcount to a mix of people and digital labor, allocated with the deliberateness once reserved for hiring. One caveat belongs on all of it: the exposure figures are sector-level classifications, not measurements of individual firms' AI use, so they describe the weather, not any one company's forecast.

The issues, by agreement

How many independent organizations name each issue as a problem. An issue is only as real as the number of separate publishers that identify it, so the count is the ranking. Bars are organizations, not documents. Where the count reads ours, no publisher here states the issue and the analysis is our own.

names it as a problemdisputes itqualifies it
A wave of new roles is arriving with no place on the chart
2
AI splits every role toward one of two tracks, and design decides which
1
The bottom rung is being removed faster than it is being redesigned
1
Headcount is no longer the planning unit
1

Who takes which position

The chart above counts positions; this shows whose they are. Read down a column for what one organization holds across the whole topic, and across a row for who lines up on one issue. Where a cell carries more than one position, the strongest is shown and the rest are in the tooltip.

Ddisputes it Qqualifies it Nnames it as a problem Pproposes a fix
Roles & the shape of the workforce: 4 issues against the 3 organizations cited on them. The number under each name is how many of these issues it is cited on.
Issue Microsoft · 4 PwC · 2 Boston Consulting Group · 1
A wave of new roles is arriving with no place on the chart N · N
AI splits every role toward one of two tracks, and design decides which P N ·
The bottom rung is being removed faster than it is being redesigned D N ·
Headcount is no longer the planning unit N · ·

A dot means this organization is not cited on that issue. It does not mean they are silent on it: an organization is cited where its document takes a position we could locate, and the absence of a citation is the absence of a finding, not a finding of absence. Who is represented lists everyone working on this topic, including those not cited above.

Where they disagree

Points where the research contradicts itself. These are the entries worth the most attention, because no single document reveals them - they only appear when the sources are read against each other.

The bottom rung is being removed faster than it is being redesigned

Entry-level work is where the technology bites first, and early-career hiring in the most exposed occupations has flatlined. The organizations still hiring at entry level are demanding senior skills from day one - without redesigning how anyone acquires them.

The majority positionJune 2026

PwC — Reports early-career postings flatlining only in the most AI-exposed quartile, and the most exposed entry roles seven times more likely to demand traditionally senior skills - with redesigned entry roles growing 35% since 2019 while others decline 10%.

The dissentApril 2025

Microsoft — Reports the identical structural change with the opposite valence, as a career accelerator rather than a squeeze - though the day-one-manager claim is made only of the roughly 9% of surveyed organizations it classes as furthest along, and no figure on entry-level hiring volumes is offered either way.

The issues in full

Each issue carries the organizations that name it, the numbers behind it, and the remedies proposed - with the concrete steps under each. Every citation points at a section of a named document, so any count here can be checked.

Issue 012 organizations name itnewest evidence Apr 2025

A wave of new roles is arriving with no place on the chart

AI trainers, agent specialists, governance leads, enablement roles - the majority of leaders say they are considering such hires, but few organizations have decided which are real jobs, where they report, or which existing roles they replace.

The lists are long and overlapping: one survey has 78% of leaders considering AI-specific roles - rising to 95% among the most advanced adopters - with agent specialists, trainers and return-on-investment analysts at the top; a platform playbook catalogues a dozen more across strategic, technical and enablement categories. The functions feel the disruption unevenly: inside one function alone, the three highest-impact roles face 55-75% skill disruption. The risk is both directions at once - titles created for fashion with no real scope, and real gaps (evaluation, agent operations) left unfilled because no existing function claims them.

How to fix it — 1 approach, 3 steps

Triage the new roles: adopt, absorb, or decline - in writing

Run the emerging-role catalogue against the organization's actual gaps, and for each: create it, fold it into an existing role, or decline it with a reason.

Done when Every AI-specific role proposed internally or seen at peers this year is marked adopted with a written scope, absorbed into a named existing role, or declined with a reason, and the declined list has a revisit date.

  1. List the AI-specific roles proposed internally or seen at peers this year.0-30 daysCHRO
  2. For each: adopt with a written scope, absorb into a named existing role, or decline.30-90 daysCHRO
  3. Revisit the declined list twice a year; real gaps reassert themselves.ongoingCHRO
The evidence — 3 documents
OrganizationDocumentPosition
Boston Consulting GroupConsultancy · July 2024HR playbookOur reading Estimates 55-75% skill disruption across the three highest-impact roles in the human resources function alone, with new governing and shaping roles added around them.Disruption inside one functionnames it
MicrosoftHyperscaler · April 20252025: The year the Frontier Firm is bornOur reading Reports 78% of leaders considering hiring for AI-specific roles - 95% among the most advanced firms - with AI trainers, data specialists, security specialists, agent specialists and ROI analysts leading the list.The hiring intention, measurednames it
MicrosoftHyperscalerThe CIO playbookOur reading Catalogues the emerging roles across three categories - strategic leadership, technical and operational, enablement and adoption - noting many did not exist a few years ago and some will evolve from existing functions rather than be hired.The role cataloguenames it

Issue 021 organization name it2026 evidence

AI splits every role toward one of two tracks, and design decides which

The same technology is professionalising some roles and democratising others - and the two tracks diverge in growth and pay. Organizations treat the direction as fate when it is substantially a role-design choice.

The mechanism is which tasks the machine takes. Where it absorbs the basic work - screening, drafting, routine analysis - the human half concentrates on judgment and the role professionalises. Where it absorbs the complex tasks and leaves operation and oversight, the role democratises and its bargaining power falls. The measured gap between tracks: postings growing 39% against 17% since 2018, and advertised pay growing 42% faster on the professionalised side since 2021. Role redesign - deciding deliberately which tasks move to the machine - is the operating-model lever that chooses a track.

How to fix it — 1 approach, 3 steps

Decide per role family which tasks move to the machine

For each major role family, decide deliberately whether the machine takes the basic tasks or the complex ones - and accept that not deciding is a decision for the weaker track.

Done when The five largest role families each show which tasks are being automated and a classification of professionalising, democratising or drifting, and one drifting family has been deliberately redesigned as the template.

  1. For the five largest role families, list which tasks are being automated today.0-30 daysCHRO
  2. Classify each family: professionalising, democratising, or drifting undecided.30-90 daysCHRO
  3. Redesign one drifting family deliberately, as the template for the rest.90-180 daysCOO
The evidence — 2 documents
OrganizationDocumentPosition
PwCConsultancy · June 2026Two futures of jobs in the AI eraOur reading Classifies 52% of advertised jobs as democratised - the technology shifting the role toward less expert tasks - and 22% as professionalised, with professionalised postings growing 39% against 17% since 2018 and their advertised salaries growing 42% faster since 2021, from analysis of over one billion job advertisements.The fork, measurednames it
MicrosoftHyperscaler · April 20252025: The year the Frontier Firm is bornOur reading Proposes the organizational form the professionalised track implies: teams assembled around outcomes rather than functions, with agents extending each person's scope.Teams formed around goalsproposes a fix

Issue 031 organization name it1 disputes it2026 evidence

The bottom rung is being removed faster than it is being redesigned

Entry-level work is where the technology bites first, and early-career hiring in the most exposed occupations has flatlined. The organizations still hiring at entry level are demanding senior skills from day one - without redesigning how anyone acquires them.

Two findings collide. Early-career postings have flatlined only in the most AI-exposed quartile of occupations - the sole quartile where that is true. And the entry roles that survive are being rebuilt upward: seven times more likely than the least exposed to demand traditionally senior skills such as strategic decision-making and stakeholder management, with the redesigned versions growing 35% since 2019 while unredesigned ones decline 10%. The old bargain - years of drudgery as training - is ending, and the replacement (deliberate onboarding, mentorship, early responsibility with support) exists mostly in documents, not org designs. A firm that stops hiring juniors has also, quietly, stopped manufacturing its seniors.

How to fix it — 1 approach, 3 steps

Rebuild the entry pathway before the pipeline empties

Redesign entry roles around judgment-with-support rather than drudgery-as-training, and measure the junior pipeline as a leading indicator, not an HR statistic.

Done when Entry-level hiring is counted against three years ago with the functions where the rung has gone named, one entry role has been redesigned around supervised judgment with explicit mentorship, and the junior pipeline reaches the executive team as a succession risk.

  1. Count entry-level hires by function against three years ago; name where the rung is gone.0-30 daysCHRO
  2. Redesign one entry role around supervised judgment work with explicit mentorship.30-90 daysCHRO
  3. Report the junior pipeline to the executive team as a succession risk.ongoingCEO
The evidence — 3 documents
OrganizationDocumentPosition
MicrosoftHyperscaler · April 20252025: The year the Frontier Firm is bornOur reading Reports the identical structural change with the opposite valence, as a career accelerator rather than a squeeze - though the day-one-manager claim is made only of the roughly 9% of surveyed organizations it classes as furthest along, and no figure on entry-level hiring volumes is offered either way.Managers from day onedisputes it
PwCConsultancy · June 2026Two futures of jobs in the AI eraOur reading Reports early-career postings flatlining only in the most AI-exposed quartile, and the most exposed entry roles seven times more likely to demand traditionally senior skills - with redesigned entry roles growing 35% since 2019 while others decline 10%.The entry-level squeeze, measurednames it
PwCConsultancy · June 2026Two futures of jobs in the AI eraOur reading Proposes redesigning onboarding, mentorship and training to accelerate development of skills like leadership and strategic decision-making that entry roles now demand on day one.Reinvent the pathway, not just the postingproposes a fix

Issue 041 organization name it2026 evidence

Headcount is no longer the planning unit

Workforce planning still counts people. The organizations furthest along are planning a mix - how many agents for which tasks, how many humans to direct them - and allocating machine capacity with the deliberateness once reserved for hiring.

Two formulations of the same shift: the human-agent ratio as a designed variable, set per role and function rather than emerging by accident; and machine capacity allocated like headcount - who gets it, how much, for what work, with the same accountability for outcomes. The survey behind the first reports 45% of leaders prioritizing expanded capacity through digital labor - alongside, not instead of, the 47% prioritizing upskilling - while a third consider headcount reductions. The economics move fast enough that the calculation cannot be annual: what a task costs to run this quarter is not what it costs next. None of the infrastructure for these decisions - allocation, accounting, review - exists in most organizations yet.

How to fix it — 1 approach, 3 steps

Plan capacity as a mix, and give the machine half an owner

Extend workforce planning to cover both kinds of capacity: the human-agent ratio per function, an owner for machine-capacity allocation, and a review cycle matched to how fast the costs move.

Done when At least one function states its current human-agent ratio, a named person allocates machine capacity against a stated accountability, and the ratio and its cost basis are reviewed quarterly rather than annually.

  1. Pick one function and state its current human-agent ratio, however rough.0-30 daysCOO
  2. Name who allocates machine capacity and against what accountability.30-90 daysCFO
  3. Review the ratio and its cost basis quarterly - the economics move too fast for annual.ongoingCFO
The evidence — 2 documents
OrganizationDocumentPosition
MicrosoftHyperscaler · April 20252025: The year the Frontier Firm is bornOur reading Introduces the human-agent ratio as something leaders must set per role and function, and reports 45% of leaders prioritizing expanded capacity through digital labor over the next 12-18 months while 33% consider headcount reductions.The ratio as a designed variablenames it
MicrosoftHyperscaler · June 2026Tokenomics is the new headcountOur reading Frames the allocation question - who gets machine capacity, how many, for what work - as headcount-style management with the same deliberateness and accountability, and notes the relevant cost comparison has become the human doing the same work, recalculated as prices fall.Machine capacity allocated like headcountnames it

Who is represented

This dossier is drawn from 15 organizations working on the subject, 3 of which are cited directly in the issues above.

Consultancy — 7

Boston Consulting Group 9 PwC 1 Accenture 2 Deloitte 2 McKinsey & Company 2 KPMG 1 Teneo 1

Institution — 3

World Economic Forum 3 The Conference Board 2 Citi 1

Hyperscaler — 3

Microsoft 4 IBM 3 Google Cloud 1

Vendor — 2

Eightfold AI 1 TechWolf 1