Agentic AI Systems Should Be Designed as Marginal Token Allocators
For AI system designers, this paper provides a unified economic framework to understand and address misallocation in agentic systems, though it is a position paper without empirical validation.
This position paper argues that agentic AI systems should be designed as marginal token allocation economies, showing that four layers (router, agent, serving stack, training pipeline) solve the same first-order condition. The framing explains why local token minimization leads to global misallocation and predicts specific failure modes.
This position paper argues that agentic AI systems should be designed and evaluated as \emph{marginal token allocation economies} rather than as text generators priced by the unit. We follow a single request -- a developer asking a coding agent to fix a failing test -- through four economic layers that today are designed in isolation: a router that decides which model answers, an agent that decides whether to plan, act, verify, or defer, a serving stack that decides how to produce each token, and a training pipeline that decides whether the trace is worth learning from. We show that all four layers are solving the \emph{same} first-order condition -- marginal benefit equals marginal cost plus latency cost plus risk cost -- with different index sets and different prices. The framing is deliberately minimal: we do not propose a complete theory of AI economics. But adopting marginal token allocation as the shared accounting object explains why systems that locally minimize tokens globally misallocate them, predicts a small set of recurring failure modes (over-routing, over-delegation, under-verification, serving congestion, stale rollouts, cache misuse), and points to a concrete research agenda in token-aware evaluation, autonomy pricing, congestion-priced serving, and risk-adjusted RL budgeting.