🤖 AI Summary
This work addresses the absence of real-time cost and carbon emission constraints in existing AI agent systems, which typically rely on post-hoc monitoring. The authors propose the first verifiable, zero-overrun budget and carbon gating mechanism embedded directly within the agent execution loop, grounded in the SARC governance framework and enforcing predictive governance at four critical decision points. Their approach integrates quantile conformal calibration, soft Lagrangian penalties, and both synthetic and real task arrival models, and they release a complete open-source governance library. Experiments on real-world multi-step tasks demonstrate zero budget overruns and achieve end-to-end reductions of 47–55% in token usage, monetary cost, and carbon emissions, with all results fully reproducible.
📝 Abstract
Agentic AI systems act through tools and sub-agents, yet the controls meant to bound their financial and environmental cost still sit on dashboards evaluated beside or after execution. Green SARC applies the SARC governance-by-architecture framework -- four enforcement sites in the agent loop -- to FinOps and GreenOps, contributing the theory of what to enforce and how to predict it. We report four policy-independent results. (i) The unconstrained "State Snowball" is $Θ(n^2)$ in loop depth; on 3,000 real multi-step plans (SWE-rebench) it holds on 100%, with median curvature $\hat{c}_2=216$ exceeding the linear-accretion prediction $p/2=134$ -- real plans accrete faster than the model. (ii) On real residuals the Normal-$σ$ gate under-covers (92% at nominal 95%); split-conformal calibration holds (95.2%). (iii) A soft Lagrangian penalty tuned to the budget in expectation breaches it on 91.5% of seeds; the architectural gate breaches 0%. (iv) Under binding budgets the gate's over-budget incidence is 0% on synthetic and real (BurstGPT) arrivals. End-to-end token/USD/carbon savings (47--55%) are real but policy-dependent in magnitude -- set by a scope-cap knob, not by gate rejections. The library is open-source, dependency-free, and ships a regeneration script for every cited number.