Ten objections to the Chief Workforce Officer, answered

An argument that has not met its objections is a slogan. Here are the ten I hear most, each stated as strongly as I can state it, with my answer. Three of them I partly concede.

1. It is another executive title, and organisations have too many. Agreed, if it is only a title. The argument of this series is for a mandate, which in most organisations should be held by the existing chief human resources officer. If the role can be filled without a new hire, and it usually can, this objection falls.

2. The CIO already owns the systems; this duplicates that. The CIO owns whether the systems work. Nobody owns what they do to the division of work between people and machines. The two are different, and the essay on the CIO in this series sets out why both are needed and how they relate.

3. Business units know their work best; central control will slow them down. Business units should go on running their work and choosing their tools. What they should not do alone is make the composition decision for the organisation, because they bear its cost and the organisation bears its consequence, in year five. The CWO sets boundaries and standards; it does not approve every subscription.

4. Verification will be automated, and then the whole premise collapses. This is the strongest objection and I have said so throughout: if automated checking drives verification cost to near zero, agents become ordinary capital and the role shrinks toward conventional workforce planning. I do not think that is where the evidence points, because automated checkers share the blind spots of the systems they check. But it is the condition under which I would abandon the argument, and I have written it down.

5. Regulation will impose accountability anyway. Regulation will require that someone be accountable. It will not say who. Every previous compliance regime created the duty and left the organisation to create the role, and the gap between the two was filled by incidents. The law presupposes the role; it does not supply it.

6. Treating agents as a "workforce" is exactly the anthropomorphism you warn against. A fair hit on the vocabulary, and the reason I use "productive capacity" where I can. But the alternative vocabularies are worse: "technology" hides the fact that agents now do what were jobs, and "tools" hides the fact that they decide. The word "workforce" at least keeps the question of responsibility in view. The lifecycle verbs, selected not hired and the rest, are there to stop the word from sliding into the thing it must not mean.

7. Small companies cannot afford this. They cannot afford the executive; they can afford the decision. A named person, a one-page register, a signature against each agent and two numbers watched. The essay on the two-hundred-person company sets this out, and the small company needs it more, not less.

8. The evidence of displacement is thin; you are building a role on a fear. Partly conceded. The realised effects are small so far and the one displacement finding is unreplicated. But the role is not built on displacement. It is built on the accountability gap, which exists today in every organisation running an agent nobody has signed for, whatever the labour-market aggregates say.

9. Moderna solved this by merging HR and IT. Moderna saw the problem early and I respect the move. Merging removes the independent voice for the workforce from the top table and settles every argument between system and people inside one function. It is the right instinct and the wrong structure, and the essay on Moderna explains why.

10. This is a consultant's framework dressed as a theory. The test is whether it makes predictions that could fail. It does: that firms under-deploy agents where verification is costly, that displacement lands first at the entry layer, and that the substitution between people and agents varies with governance rather than only with technology. The Institute behind this series exists to test those claims and to publish the result whichever way it goes. A framework does not include its own kill criteria. This one does.

Three partial concessions out of ten is, I think, about right for an argument that is new. The remaining seven are where the organisation's attention should go, and the sooner the better, because the agents are already at work.