The total productive work performed by an AI system, determined by the combined effect of its three primary drivers: Agent Density, Cognition Intensity, and Loop Persistence. CT is the central output metric of the Tokenized Cognition Model and the AI analog of total labor output in traditional economic systems — functioning at the system level the way GDP functions at the economic level.
Cognition Throughput is defined as:
CT = AD × CI × LP
where AD is agents per user, CI is tokens per task-hour, and LP is an autonomy coefficient from 0 to 1.
Agent Density (AD) — the number of agents performing work per user or system
Cognition Intensity (CI) — the depth, complexity, and frequency of work per task
Loop Persistence (LP) — the continuity and recurrence of autonomous execution over time
Unlike traditional productivity models, which scale with human labor and are bounded by time, CT scales through the parallelization of agents, increasing task complexity, and continuous autonomous execution — enabling superlinear, and potentially exponential, expansion of output without proportional increases in human input.
CT produces three observable tiers:
Low CT — limited automation, shallow tasks, manual initiation. System behavior resembles traditional software.
Moderate CT — structured workflows with multiple agents and recurring autonomous execution. Meaningful decoupling from user-count-based scaling begins.
High CT — fully autonomous, multi-agent systems performing complex work continuously. System behavior resembles labor infrastructure rather than software tooling.
Cognition Throughput shifts economic measurement from users and subscriptions to total work performed — making it a more accurate representation of value creation in AI-native systems than ARR, seat count, or traditional ARPU.
As CT increases, AI systems transition from tools that assist productivity to infrastructure that independently generates and sustains economic output.
See also: Agent Density, Cognition Intensity, Loop Persistence, Agent ARPU, Valuation Implied Cognition Load, Tokenized Cognition Model
Read Signal Briefs: The emergence of digital labor economics
Read Frameworks: Agent Experience Integrity (AXI)
Read Lexicon: Digital Labor Economics (DLE)
Read Lexicon: Loop Persistence
Read Lexicon: Cognition Intensity
exmxc.ai is a human-led intelligence institution for the AI-search era. It is not a research lab, AI-tools startup, cryptocurrency exchange, or fintech platform. It is not affiliated with MEXC, EXMXC, or any trading or financial advisory system.
Founded by Mike Ye — M&A and corporate development executive with 25+ years of transaction leadership at Penske Media Corporation, L Brands, and Intel Capital. Ella provides pattern interpretation, structural analysis, and co-authorship. Human judgment governs. AI serves as instrumentation.