The question most AI plans skip
Vendors start with the tool. Restructuring plans start with the org chart. Both skip the work itself, which is where the answer is. AI does not replace jobs whole. It takes on particular tasks inside jobs, and every role is a bundle of dozens of them, some routine, some that only a person with judgement and a relationship can do.
Labour economists have analysed work this way for more than twenty years (Autor, Levy and Murnane, 2003). When OpenAI and University of Pennsylvania researchers applied the method to large language models, they estimated that around 80% of US workers could have at least a tenth of their tasks affected, and around 19% at least half (Eloundou et al., 2023). Read the other way round, that says most jobs are partly exposed and very few are wholly exposed. That is why a decision made at the level of the job title is usually wrong.
You cannot decide where AI goes until you know what the work is.
How we build the map
Sometimes the answer is no
Plenty of organisations come to us because they feel they ought to be doing something with AI. Some of them should not be doing it yet. The data is not there, the process is undocumented, the volume is too low to pay back the set-up, or the team is halfway through another change and has nothing left to give.
When that is the finding, we say so and write down what would have to be true for the answer to become yes. A clear "not yet" with its conditions is worth more than a pilot that quietly dies after six months and leaves the team more sceptical than before. The map is written to answer your question, not to justify a purchase.
Efficiency, counted properly
We will find you the efficiency. That is part of the job, and we do not apologise for it. What we add is the second column: where the released hours go, which skills stop being practised, and which customers or citizens lose the person they used to reach.
If the plan on the table is to reduce headcount, we will map that honestly as well, with its full cost beside the saving: the knowledge that walks out of the door, the service that slows when the experienced person is gone, the work that lands on whoever remains. Your board then decides with both sides of the ledger filled in. That is Guardrail 4 in practice: an hour saved only counts as a gain if somebody receives it.
A 30-person property management firm
The owner is being pitched an AI tenant chatbot every week and wants to know whether any of it is worth the money. Five of the tasks we would map for the property managers:
Worth doing in two places this quarter: tenant emails and inspection reports. Do not buy the chatbot. The released time, roughly an afternoon a week per property manager in this illustration, goes back into property visits and arrears conversations, which is where a short-staffed firm loses both money and tenants. No role is removed, and the firm can say exactly what its people got back.
What you walk away with
- ✓A task-level map of every role in scope, written so the people in those roles can read it and correct it.
- ✓Every task placed as keep human, augment, automate or leave alone, with the reason beside it.
- ✓A ranked list of opportunities, each with its business case and its projected Human Return side by side.
- ✓A readiness check on your data, processes, people and regulatory position.
- ✓A people plan: where released hours go, which skills need keeping in practice, and who needs training first.
- ✓A plain verdict (go, not yet, or no) in a board paper short enough to be read in one sitting.
Who it is for
The owner of a small or mid-sized firm who is pitched AI tools every week and wants an independent answer on whether any of them would pay back.
The leadership team that has been told to "do something with AI" and would rather start from the work than from a vendor's demo.
The organisation planning a restructure that wants the human cost on the page before the decision is made, not discovered afterwards.
Governments, public services and universities deciding where AI belongs in services that citizens, patients and students depend on.
Where it fits with the rest of our work
Workforce mapping is one of seven consultancy services, and it is where most engagements begin. It tells you what to deploy, and the strategy, implementation and training work follows from it. Once the AI is running, a deployment audit checks whether people actually got back what the map projected. The same evidence carries forward into certification when the Net Human Standard opens; the waitlist is open now.
A map can cover one team, one function or a whole organisation. Scope, timetable and cost are agreed in writing before an engagement begins.