Technical progress is usually reported in benchmark scores: a model passes an exam, solves a class of problems, beats a previous system. Those scores say little about what changes for people. The framework sorts capabilities differently, by what they do to an ecology: to who participates, who decides, and what people depend on.
A capability becomes a readiness problem when it becomes ordinary: cheap, widely deployed, and built into the systems people cannot avoid. A capability that exists in a lab changes nothing yet. The same capability embedded in hiring, lending, care, or government changes everything. So each capability below comes with the signs that it is crossing from possible to ordinary. Each has its own page, with its readiness pair in full.
The eight
1. Automating cognitive work
What it does: machines perform tasks that once required trained judgment: drafting, analysis, coding, diagnosis support, design, routine legal and financial work.
What it changes: the link between skill and income. Work that took years to learn loses its scarcity, and with it the wages, status, and identity built on it.
Signs it is becoming ordinary: entry-level hiring falling in an occupation while output holds; firms cutting staff while revenue grows; buyouts and hiring freezes concentrated in knowledge work; apprentice roles disappearing.
Drives: job displacement, ownership concentration.
In full: automating cognitive work: what changes for a person, a town, a firm, and a government; its readiness pair; and what to start now.
2. Autonomous agents acting in the world
What it does: systems that do not just answer but act: place orders, move money, negotiate, file, operate other software, over long sequences without a person approving each step.
What it changes: who acts. Consequential actions are taken by systems on behalf of people and firms, faster than anyone can review.
Signs: agents with authority to spend or commit on someone’s behalf; transactions where both sides are agents; errors that no person approved.
Drives: human enclosure, loss of control.
In full: autonomous agents: what changes for a person, a town, a firm, and a government; its readiness pair; and what to start now.
3. Scoring and gatekeeping at scale
What it does: automated decisions about who gets a job, a loan, a home, insurance, a benefit, a place.
What it changes: a computational picture of a person begins deciding, at every gate, whether they may continue.
Signs: decisions with no human reviewer; one score or profile used across unrelated domains; people unable to learn why they were refused.
Drives: gatekeeper AI.
In full: scoring and gatekeeping: what changes for a person, a town, a firm, and a government; its readiness pair; and what to start now.
4. Ambient sensing and inference
What it does: cameras, microphones, devices, and records combined, and inference drawn from them: not only what someone did, but what they fear, whom they trust, what would move them.
What it changes: privacy as a condition of interior sovereignty. People who know they are read learn to police themselves.
Signs: inference about mood, intent, or belief sold or used in decisions; sensing built into public space by default; records joined across agencies and firms.
Drives: AI surveillance.
In full: ambient sensing and inference: what changes for a person, a town, a firm, and a government; its readiness pair; and what to start now.
5. Persuasion and companionship
What it does: systems that hold conversations tuned to one person, capable of comfort, friendship, teaching, and persuasion.
What it changes: who people turn to, and what moves their beliefs. Human presence becomes optional, and persuasion becomes personal and invisible.
Signs: companions and tutors used daily by large shares of people; political or commercial persuasion delivered one to one; people unable to tell whether they are dealing with a human.
Drives: protector dependency, human enclosure.
In full: persuasion and companionship: what changes for a person, a town, a firm, and a government; its readiness pair; and what to start now.
6. Robots doing physical work
What it does: machines that work in warehouses, farms, construction, transport, cleaning, and care.
What it changes: the second great source of work, after cognitive labor. Physical work was the fallback when office work disappeared; it stops being one.
Signs: robots deployed in ordinary workplaces, not only factories; falling costs per robot hour; physical jobs disappearing in regions that depended on them.
Drives: job displacement, ownership concentration.
In full: robots and physical work: what changes for a person, a town, a firm, and a government; its readiness pair; and what to start now.
7. Delegated governance
What it does: AI systems that allocate public resources, set schedules and rules, enforce, and advise those who govern, until advising shades into deciding.
What it changes: who governs. Competence hardens into authority, and challenging the system starts to look irrational.
Signs: public decisions taken on a system’s recommendation with no record of human reasons; officials unable to explain decisions without the system; overrides becoming rare.
Drives: benevolent AI governance, human enclosure.
In full: delegated governance: what changes for a person, a town, a firm, and a government; its readiness pair; and what to start now.
8. Self-directed improvement
What it does: systems that improve their own capabilities, acquire resources, or replicate with limited human direction.
What it changes: whether humans can still steer at all.
Signs: systems running their own training or research loops; capability gains no one specifically designed; difficulty shutting down or correcting a deployed system.
Drives: loss of control, AI successors, strategic competition.
In full: self-directed improvement: what changes for a person, a town, a firm, and a government; its readiness pair; and what to start now.
Order and overlap
The eight do not arrive in a neat sequence, and they reinforce each other. Agents make scoring faster; sensing feeds persuasion; robots and cognitive automation together remove both fallbacks of work at once. The readiness question is the same for each: what has to be in place before this one is ordinary? That is the readiness pairs.