Applied Frontier

Pilots · Last updated November 2026

How a targeted pilot answers a question about the AI economy

Direct answer

A targeted pilot is how Applied Frontier answers a question about work in the AI economy: a hypothesis about a new job family, a short program that creates that job for real people in one real market, and a measured cost per job created, time to placement and retention. It is the difference between a research venture and a venture lab.

Why pilots and not reports?

Reports about the future of work describe job families that may or may not exist. A pilot creates one. By the end, a ministry or a multilateral agency knows what the role is, who can be trained into it, how long that takes, what it costs, and whether employers keep the people. That evidence is what makes a workforce program fundable.

What worked for the past 200 years will not work for the next five, and nobody has run the experiments yet. We run them.

What does a pilot look like?

Stage What happens Typical length
HypothesisOne job family, one market, one claim about who can be trained into it and what it will pay2 weeks
Market and employersSecure the employers who will hire, agree the demonstrated-work standard they will accept3–4 weeks
ProgramTrain a cohort in the role, priced in weeks and hundreds of dollars, with AI tools doing most of the production work6–12 weeks
MeasureCost per job created, time to placement, starting wage, employer acceptance, retention at 6 and 12 monthsThrough month 12
PublishThe result, the sources and the method, dated and revised when the models changeAt month 3 and month 12

The kind of question a pilot answers

Where to run a task

Which market on earth is the best place to run a data-annotation task in Arabic?

Where to build

Where is the best place to build a data center, judged by the talent available and not only the capital?

What an industry needs

What new jobs are needed to double the output of the aerospace industry, and who can fill them?

What a new model needs

What job families would a drone-based defense model require, and which countries are best equipped to build them?

The same method answers a ministry’s question about its own population and a company’s question about where to put its next facility.

What we measure

Cost per job

total spend divided by verified placements

Weeks to placement

from program entry to paid work

Retention

still employed at 6 and 12 months

Plus starting wage against the local and global rate for the role, and whether the employer accepted the demonstrated-work credential in place of a degree. Every figure is published with its method.

How to commission a pilot

A pilot needs three things from a partner: a market, a budget line, and access to employers or hiring data. Ministries, multilateral agencies and foundations can commission one directly; companies with a siting or workforce question can commission one privately.

Work with us

Questions about pilots

How long does a targeted pilot take?
Twelve to twenty weeks from hypothesis to first placements, with retention measured at six and twelve months afterward.
Who pays for a pilot?
A ministry, a multilateral agency or a foundation, usually on partly outcomes-based terms where part of the fee is tied to verified placements.
What if the pilot fails?
That is a result. A measured cost per job that is too high, or a job family that does not retain, is published alongside the successes; the point is evidence, not advocacy.