Compounding
A practice for senior operators in the AI age
Compounding is the practice of holding judgement, relationships, and accumulated context steady while the labour market re-prices around the operator. It is calibrated for the senior operator — the person whose value to an organisation lives in twenty thousand meetings of pattern recognition, four phone calls that get returned, and a decade of decisions whose reasons were never written down.
That value is currently exposed to two structural failures.
The first is departure. When the senior operator leaves, all of it disappears with them. None of it is in the handover document. None of it could be. It was never written down, because the operator carrying it did not know they were carrying it. That is how senior expertise works. It accumulates invisibly, gets rewarded for being invisible, and walks out the door in a single afternoon.
The second is narrowing. When the senior operator stays under sustained pressure, judgement contracts. The brain under chronic threat optimises for survival; the aperture closes; the operator becomes a worse version of themselves while believing they are simply working harder. Performance psychology has documented this for forty years. The science is settled. The infrastructure to address it has not been built.
Both failures produce the same outcome. The value the organisation depends on becomes unavailable. There is no backup.
The macro condition makes this acute rather than chronic. Technology is structurally deflationary. The post-Keynesian economic system runs on inflation. For most of the twentieth century, productivity gains were absorbed by wage growth and consumption expansion; the absorption mechanism worked. AI breaks the absorption mechanism by deflating cognitive labour itself — the labour the senior operator was paid for, and built a life around. The senior operator is caught in the scissors: depreciating expertise, appreciating cost-of-living. This is structural, not cyclical. It does not resolve.
Compounding inherits from performance psychology rather than competing with it. The psychology trains the operator's mind. Compounding trains the system around the mind. Each is half a solution; together they are sufficient.
The practice has four layers. Capture compounds context — what the operator knows about their own world. Watch compounds vigilance — what the operator is alerted to. Reason compounds insight — what the operator can synthesise across time. Act compounds capacity — what the operator can do without depleting themselves. A cognitive load principle runs through all four: the system reduces operator load rather than adding to it. This is what distinguishes Compounding from automation.
The economics are what make the practice defensible. Most AI spend depreciates: tokens spent, value realised once, then gone. Compounding spend appreciates: each token spent on capture and structure increases the value of every prior and future token. The mathematics works the same way as financial compounding, because it is the same mathematics. This makes Compounding the only operator-side AI practice that produces appreciating returns on token spend.
Most AI spend depreciates. Compounding is the practice that makes it appreciate.
The opening essay is the long argument. The manifesto follows. The book follows that. For consultancy enquiries or to discuss the work, write to paul@paulroebuck.com.