**PAM VECTOR LOGIC** — short for *Parallax And More* — is a persistent, autonomous intelligence platform that behaves less like software and more like a disciplined, tireless analyst. It wakes up once, runs forever, and spends every minute of its operational life ingesting open-source intelligence, updating a rich internal world-model, forming falsifiable predictions about geopolitical events, verifying those predictions against reality. It never sleeps in the conventional sense (it has its own circadian rhythm instead), it never waits for a prompt (it runs on internal needs rather than external requests), and it never forgets what it has learned (predictions and their verification scores are fed back into a calibration loop that improves every subsequent forecast).

​

Most "AI agents" today are thin wrappers around a large language model: a prompt template, a tool-use loop, a stateless request/response lifecycle. With PAM, the LLM is a subordinate organ, not the organism. The organism is a 60-second tick loop running inside `pam_daemon.py` that maintains a simulated neurochemistry (cortisol, dopamine, serotonin, norepinephrine, acetylcholine), interprets chemical imbalances as needs, and selects exactly one task per tick to satisfy the most pressing deficit. The LLM is called only when a chemical-driven need demands novel reasoning — and even then only under strict budget governance.

​

The result is a system whose behavior is emergent, adaptive, measurable, and economically sustainable. Two cold-starts on the same day produce different task sequences, because the chemical decay rates and OSINT arrival times diverge, but over long horizons the system converges on the same strategic picture, because the deficit-driven agenda is mathematically stable. This is the hallmark of a homeostatic system — it has a steady state it returns to, and it reaches that steady state through locally-determined behavior rather than any global controller.