2026-11-15T17:42:00Z https://andygrove.systems/oai

oai:andygrove.systems:monograph-20262026-11-15

High Output Operations for Human + Agent Teams: Limiting Steps, Queueing & L0–L3 Task-Relevant Maturity (2026 Monograph & 42-Hour CPD Syllabus) Andygrove.systems Industrial Engineering & Operations Research Collective Continuing Professional Development; Fresh AI Agent Preprints; Little's Law; Kingman Queueing; Task-Relevant Maturity Open-access Continuing Professional Development (42 PDH / CPE / CEU / CPD hours) and Executive Upskilling Hub built directly on 16 fresh 2023–2026 arXiv, NBER & HBS preprints (https://andygrove.systems/preprints) and 6 quantitative self-study modules (https://andygrove.systems/education) for Chief Operating Officers (COOs), Industrial Engineers, Operations Researchers, BizOps & Program Managers, and Process & Quality Engineers. 2026-11-15 Text application/pdf https://andygrove.systems/papers/andygrove-hybrid-operations-monograph-2026.pdf https://andygrove.systems/preprints https://andygrove.systems/education en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-012026-11-15

Queue & AI: When Faster Tasks Slow Down the Workflow — Continuing Professional Upskilling Study Brief (arXiv:2605.27202) Silvia Bartolucci, Pierpaolo Vivo (UCL & King's College London) Professional Upskilling; QUEUEING & VARIANCE WEDGE; COO Operations Proves how fast AI draft generation creates a 'spike-and-tail' service distribution ('variance wedge'): escaped errors return as heavy rework while congested human reviewers rationally lower scrutiny precisely when risk peaks. COO Production Rule: Never scale agent draft volume without verifying t_rev + p_err·t_fix < t_manual, α_AI > α_crit, and filtering defects before human reviewers. 2026-11-15 Text https://andygrove.systems/preprint/pp-01 https://andygrove.systems/papers/pp-01.pdf https://arxiv.org/abs/2605.27202 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-022026-11-15

AgentRouter: Heterogeneous Model Routing for Cost-Optimal Multi-Step Agentic Workflows — Continuing Professional Upskilling Study Brief (arXiv:2609.22951) Rudrendu Kumar Paul, Sourav Nandy Professional Upskilling; STEP-LEVEL COST ROUTING; COO Operations Trains a <5ms 12M-parameter trajectory router across 50,000 steps (GAIA, SWE-bench, WebArena, ToolBench) and 4 model tiers, cutting inference cost by ~72% while retaining 97.3% of frontier-only performance (vs. 31% for single-turn RouteLLM). COO Production Rule: Route per trajectory step (parsing/tool-formatting → small tier; ambiguous synthesis → frontier tier), never one frontier model per whole workflow. 2026-11-15 Text https://andygrove.systems/preprint/pp-02 https://andygrove.systems/papers/pp-02.pdf https://arxiv.org/abs/2609.22951 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-032026-11-15

Single-Agent LLMs Outperform Multi-Agent Systems on Multi-Hop Reasoning Under Equal Thinking Token Budgets — Continuing Professional Upskilling Study Brief (arXiv:2604.02460) Dat Tran, Douwe Kiela (Stanford University) Professional Upskilling; TOKEN BUDGET & TOPOLOGY; COO Operations Uses the Data Processing Inequality (DPI) to prove that under equal thinking-token budgets, multi-agent message passing loses mutual information and trails single-agent reasoning unless context rot on noisy long inputs degrades single-agent focus. COO Production Rule: Audit every multi-agent architecture at equal token budget; eliminate decorative inter-agent chatter unless noisy-context compression or privilege isolation requires it. 2026-11-15 Text https://andygrove.systems/preprint/pp-03 https://andygrove.systems/papers/pp-03.pdf https://arxiv.org/abs/2604.02460 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-042026-11-15

When Should AI Agents Ask? Clarification Requests Under Limited Human Attention — Continuing Professional Upskilling Study Brief (arXiv:2610.41129) Hanzhao Wang, Shengjie Liu, Zhengchao Wang Professional Upskilling; HUMAN ATTENTION QUEUES; COO Operations Models parallel AI agents sharing one human supervisor with finite attention as a queueing dynamic program; introduces Pairwise Coordination Policy (PCP) so agents condition escalation on real-time human queue congestion. COO Production Rule: Make agent escalation thresholds queue-aware: when senior human WIP rises, throttle low-value clarifications and defer non-urgent subtasks. 2026-11-15 Text https://andygrove.systems/preprint/pp-04 https://andygrove.systems/papers/pp-04.pdf https://arxiv.org/abs/2610.41129 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-052026-11-15

CascadeDebate: Multi-Agent Deliberation for Cost-Aware LLM Cascades — Continuing Professional Upskilling Study Brief (arXiv:2604.12262) Raeyoung Chang, Dongwook Kwon, Jisoo Lee, Nikhil Verma Professional Upskilling; ESCALATION CASCADES; COO Operations Injects lightweight multi-agent deliberation strictly at uncertain escalation boundaries of a cost-aware LLM cascade, resolving borderline ambiguity on small models before paying frontier-tier or human escalation costs. COO Production Rule: Trigger consensus/verification ensembles only on borderline-confidence items, keeping clear routine traffic single-pass. 2026-11-15 Text https://andygrove.systems/preprint/pp-05 https://andygrove.systems/papers/pp-05.pdf https://arxiv.org/abs/2604.12262 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-062026-11-15

AI Agents for Inventory Control: Human-LLM-OR Complementarity — Continuing Professional Upskilling Study Brief (arXiv:2602.12631) Jackie Baek, Yaopeng Fu, Will Ma, Tianyi Peng Professional Upskilling; HUMAN + LLM + OR TRIAD; COO Operations Evaluates multi-period operational control on InventoryBench, proving that pairing LLMs (unstructured context translation) + exact Operations Research solvers (constrained stochastic math) + human oversight beats standalone LLMs or pure OR. COO Production Rule: Never pay reasoning tokens to approximate linear/stochastic optimization; let LLMs parameterize exact OR solvers under human governance. 2026-11-15 Text https://andygrove.systems/preprint/pp-06 https://andygrove.systems/papers/pp-06.pdf https://arxiv.org/abs/2602.12631 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-072026-11-15

Intelligent AI Delegation — Continuing Professional Upskilling Study Brief (arXiv:2602.11865) Nenad Tomašev, Matija Franklin, Simon Osindero (Google DeepMind) Professional Upskilling; DELEGATION GOVERNANCE; COO Operations Establishes a formal operational framework for multi-hop human-to-agent and agent-to-agent delegation covering dynamic authority bounds, contract specification, verification checkpoints, and accountability chains. COO Production Rule: Assign one named human owner and pair every delegated agent permission with a machine-verifiable acceptance test and rollback switch. 2026-11-15 Text https://andygrove.systems/preprint/pp-07 https://andygrove.systems/papers/pp-07.pdf https://arxiv.org/abs/2602.11865 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-082026-11-15

DelegationBench: Measuring When AI Agents Should Ask Before Acting — Continuing Professional Upskilling Study Brief (arXiv:2610.05532) Shiva Pochampally Professional Upskilling; ABSTENTION & ESCALATION EVALS; COO Operations Tests 10 models across 156 matched-pair irreversible vs. reversible scenarios: surface prompt rephrasing swings act-now rates by up to 52.5 percentage points, and live tool-executing agents check in far less than static evaluators. COO Production Rule: Enforce hard programmatic L1 gates (not prompt politeness) before high-blast-radius actions (external emails, refunds, production DB writes). 2026-11-15 Text https://andygrove.systems/preprint/pp-08 https://andygrove.systems/papers/pp-08.pdf https://arxiv.org/abs/2610.05532 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-092026-11-15

Why Do Multi-Agent LLM Systems Fail? (MAST Taxonomy) — Continuing Professional Upskilling Study Brief (arXiv:2503.13657) Mert Cemri, Melissa Z. Pan, Shuyi Yang, Lakshya A. Agrawal, Matei Zaharia, Ion Stoica et al. Professional Upskilling; MULTI-AGENT FAILURE TAXONOMY; COO Operations Analyzes 1,600+ execution traces across 7 multi-agent frameworks, classifying 14 failure modes across specification flaws, inter-agent misalignment, and absent task verification. COO Production Rule: Institutionalize weekly trace 1-on-1s: most multi-agent failures are handoff contract and inspection-gate defects, not base-model IQ limits. 2026-11-15 Text https://andygrove.systems/preprint/pp-09 https://andygrove.systems/papers/pp-09.pdf https://arxiv.org/abs/2503.13657 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-102026-11-15

The Cybernetic Teammate: A Field Experiment on Generative AI Reshaping Teamwork and Expertise — Continuing Professional Upskilling Study Brief (NBER w33641) Fabrizio Dell’Acqua, Charles Ayoubi, Raffaella Sadun, Ethan Mollick, Karim R. Lakhani et al. Professional Upskilling; HYBRID TEAM ORG DESIGN; COO Operations Pre-registered field experiment with 776 P&G professionals: 1 human + AI matched the performance of 2-person cross-functional human teams while erasing R&D vs. commercial functional silos. COO Production Rule: Redesign bloated cross-functional handoff chains into compact '1 generalist-specialist + AI' pods, reserving multi-human review for final sign-off. 2026-11-15 Text https://andygrove.systems/preprint/pp-10 https://andygrove.systems/papers/pp-10.pdf https://www.nber.org/papers/w33641 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-112026-11-15

Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters — Continuing Professional Upskilling Study Brief (arXiv:2408.03314) Charlie Snell, Jaehoon Lee, Kelvin Xu, Aviral Kumar Professional Upskilling; TEST-TIME COMPUTE ALLOCATION; COO Operations Shows that allocating inference compute adaptively by prompt difficulty via process verifiers lets smaller models beat 14× larger models at substantially lower total FLOPs. COO Production Rule: Allocate extra verification tokens strictly to hard tail tasks rather than paying frontier parameter prices on 80% routine volume. 2026-11-15 Text https://andygrove.systems/preprint/pp-11 https://andygrove.systems/papers/pp-11.pdf https://arxiv.org/abs/2408.03314 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-122026-11-15

AI Agents That Matter — Continuing Professional Upskilling Study Brief (arXiv:2407.01502) Sayash Kapoor, Benedikt Stroebl, Zachary S. Siegel, Nitya Nadgir, Arvind Narayanan Professional Upskilling; PARETO COST-ACCURACY EVALS; COO Operations Demonstrates that SOTA agent scaffolds often burn 10×–50× higher API cost for negligible accuracy gains and fail on holdout robustness when evaluated without joint cost-accuracy Pareto curves. COO Production Rule: Mandate 2D Pareto evals (Verified Accuracy vs. USD/Task) before promoting any agent scaffold from R&D into production. 2026-11-15 Text https://andygrove.systems/preprint/pp-12 https://andygrove.systems/papers/pp-12.pdf https://arxiv.org/abs/2407.01502 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-132026-11-15

RouteLLM: Learning to Route LLMs with Preference Data — Continuing Professional Upskilling Study Brief (arXiv:2406.18665) Isaac Ong, Amjad Almahairi, Wei-Lin Chiang, Joseph E. Gonzalez, Ion Stoica et al. Professional Upskilling; PREFERENCE-TRAINED ROUTERS; COO Operations Trains lightweight query routers on human preference data to dynamically select between strong and weak LLMs, cutting single-turn frontier call costs by 50–85% without degrading response quality. COO Production Rule: Deploy calibrated preference routers in front of single-turn LLM gateways and step-aware routers (PP-02) for multi-step agent loops. 2026-11-15 Text https://andygrove.systems/preprint/pp-13 https://andygrove.systems/papers/pp-13.pdf https://arxiv.org/abs/2406.18665 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-142026-11-15

When Combinations of Humans and AI Are Useful: A Systematic Review and Meta-Analysis — Continuing Professional Upskilling Study Brief (arXiv:2405.06087) Michelle Vaccaro, Abdullah Almaatouq, Thomas W. Malone (MIT) Professional Upskilling; HUMAN-AI COMPLEMENTARITY META-LAW; COO Operations Meta-analysis of 106 experimental studies (370 effect sizes): naive human+AI pairing underperforms the best solo actor on pure prediction/decision tasks, but achieves true synergy in modular creation + structured verification workflows. COO Production Rule: Do not ask humans to 'eyeball' calibrated probability outputs; partition creative synthesis, exact solver checks, and exception judgment cleanly. 2026-11-15 Text https://andygrove.systems/preprint/pp-14 https://andygrove.systems/papers/pp-14.pdf https://arxiv.org/abs/2405.06087 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-152026-11-15

FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance — Continuing Professional Upskilling Study Brief (arXiv:2305.05176) Lingjiao Chen, Matei Zaharia, James Zou (Stanford University) Professional Upskilling; LLM CASCADES & CACHING; COO Operations Introduces the foundational three-lever inference cost stack — prompt compaction, semantic/prefix approximation caching, and learned LLM cascades — matching GPT-4 accuracy with up to 98% cost reduction. COO Production Rule: Stack immutable prefix caching (EQ-10) ahead of learned confidence cascades so routine volume never hits frontier billing. 2026-11-15 Text https://andygrove.systems/preprint/pp-15 https://andygrove.systems/papers/pp-15.pdf https://arxiv.org/abs/2305.05176 en https://creativecommons.org/licenses/by/4.0/

oai:andygrove.systems:pp-162026-11-15

Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality — Continuing Professional Upskilling Study Brief (SSRN 4573321) Fabrizio Dell’Acqua, Edward McFowland III, Ethan R. Mollick, Karim R. Lakhani et al. (758 BCG Consultants) Professional Upskilling; JAGGED FRONTIER & CENTAUR SPLIT; COO Operations Inside the capability frontier, AI lifted output +12.2%, speed +25.1%, and quality +40%; on deceptive tasks just outside the frontier, AI users dropped −19 percentage points below unassisted humans due to unverified reliance. COO Production Rule: Re-map your workflow's Jagged Frontier at every model-release inflection point and enforce 'Centaur' hard task boundaries. 2026-11-15 Text https://andygrove.systems/preprint/pp-16 https://andygrove.systems/papers/pp-16.pdf https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4573321 en https://creativecommons.org/licenses/by/4.0/