Pith. sign in

REVIEW 2 major objections 6 minor 186 references

Optimal buying agents use threshold policies that change with how much they know about future prices—and work better when language models pick the model, not the buy button.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-11 14:48 UTC pith:RMFXFGMV

load-bearing objection Clean buyer-side policy menu (ODE thresholds, Bayesian learning, robust randomized thresholds) with a usable Keepa evaluation and a practical LLM-as-selector lesson. the 2 major comments →

arxiv 2607.04708 v1 pith:RMFXFGMV submitted 2026-07-06 econ.TH cs.AIcs.CYcs.GTcs.HC

Strategic Buying Agents

Mingyang Fu , Ming Hu This is my paper
classification econ.TH cs.AIcs.CYcs.GTcs.HC
keywords agentic commerceoptimal stoppingpurchase timingthreshold policiesBayesian learningcompetitive analysisminimax regretLLM agents
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

As AI agents start shopping for people, the hard part is not browsing product pages but deciding when the current price is good enough to buy before a deadline. This paper shows that the right rule depends on how much the agent knows about future prices. If price changes arrive at a known rate from a known distribution, the agent should buy when price falls below a time-dependent threshold governed by a simple differential equation. If the distribution is unknown, the same idea holds but the threshold moves with the agent’s updated beliefs. If almost nothing is known except price bounds, randomized thresholds give the best worst-case guarantees. On a thousand real Amazon shopping windows, the calibrated stationary and Bayesian rules deliver competitive average surplus, the robust rule protects the lower tail, and language models help most by choosing which rule and which history to use—not by calling buy or wait themselves.

Core claim

Across three information regimes, the optimal (or robustly optimal) purchase policy is a threshold rule: a dynamic ODE-governed threshold when the price-adjustment law is known, a belief-dependent threshold under Bayesian learning of that law, and a randomized static threshold that attains the optimal competitive ratio and minimax regret when only price bounds are trusted. On real Amazon price histories these operations-research policies, once calibrated, are competitive with simple baselines, and language models are more useful for regime and sample selection than for direct timing decisions.

What carries the argument

The purchase-threshold policy (dynamic ODE threshold in the stationary case; belief-dependent threshold under Bayesian updating; randomized static threshold under only price bounds). It converts the finite-horizon stopping problem into an implementable buy-or-wait rule whose form is dictated by the agent’s information about future prices.

Load-bearing premise

The paper leans on the idea that price changes arrive like random Poisson events from a stable distribution that can be estimated from past data—an assumption real marketplace paths often violate with bursts, calendar effects, and seller reactions.

What would settle it

On a large out-of-sample set of shopping windows where price changes are clearly non-Poisson or non-stationary, check whether the calibrated stationary and Bayesian thresholds still beat historical-min and buy-last rules on mean normalized surplus and whether the robust randomized threshold still leads at the 10th percentile.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 6 minor

Summary. The paper designs strategic buying agents that decide when to purchase a single item within a finite shopping window under three information regimes. In the stationary regime (Poisson adjustments, known H), the optimal policy is a time-dependent threshold characterized by an ODE (Prop. 1). In the Bayesian regime (known λ, unknown H with prior), the optimum remains a belief-dependent threshold (Prop. 3), with short- and long-window bounds on the information gap (Lemma 3, Prop. 6). In the robust regime (price bounds only), randomized static thresholds attain the optimal competitive ratio ρ(v) and minimax regret R̄(v) (Props. 7–8). The authors treat these rules as an implementable policy menu, evaluate them on 1,000 Keepa Amazon instances (367 items, 48,933 observations), and compare fixed OR policies, a heuristic selector, LLM-OR (LLM selects regime/samples; OR decides), and simple/LLM baselines. Stationary and Bayesian perform competitively on mean normalized surplus; robust is strongest at the 10th percentile; LLMs appear better at regime/sample selection than at direct buy-or-wait decisions.

Significance. The contribution is timely and well-scoped: as agentic commerce moves from search to delegated purchasing, the paper supplies a buyer-side OR framework that maps information structure to implementable stopping rules, rather than treating strategic buyers as a black box for seller models. Strengths include complete characterizations with full e-companion proofs (variational inequalities, ODE thresholds, MLR comparative statics, matching competitive-ratio and minimax-regret bounds), a clean hierarchy from calibrated stochastic models to robust guarantees, and an out-of-sample Keepa evaluation that tests stylized policies against real price paths. The practical finding that LLMs are more useful for regime/sample selection than for sequential buy-or-wait control is actionable for agent design. The Poisson–Calvo abstraction is stylized, but the robust regime and empirical results mitigate that limitation for the paper’s central claims.

major comments (2)
  1. Section 7.2–7.3 and Table 2: the Heuristic Selector win rate (26.5%) and opt-hit rate (21.6%) are far below fixed Stationary/Bayesian (~48%) and even below Buy Last. The abstract and introduction present the three regimes as a “policy menu for implementation,” but the only non-LLM automated selector underperforms simple baselines. Either strengthen the selection layer (or report why the fixed diagnostic cutoffs fail on Keepa) or qualify the implementation claim: the menu is valuable when selection is done by an LLM or by a better diagnostic rule, not by the current heuristic alone.
  2. Section 7.1: valuations are synthetic multipliers of the window-start price drawn from [1.01, 1.50]. This construction mechanically places v above p0 and may favor waiting policies relative to real budgets that can lie near or below recent prices. A sensitivity check with valuations calibrated to historical purchase-relevant levels (or a wider multiplier range including v near p0) would make the surplus rankings in Table 2 and Figure 7 more credible for the delegated-shopping interpretation.
minor comments (6)
  1. Assumption 1 and Section 4: briefly discuss how λ̂ is estimated under bursty or calendar-driven Amazon adjustments, and whether misspecification of the Poisson intensity systematically biases the ODE threshold upward or downward.
  2. Figure 8 note already flags factual errors in raw LLM purchase notifications; consider either filtering those outputs or treating them as a documented failure mode of free-form LLM messaging rather than leaving them as red-box asides.
  3. Proposition 6 / Assumption 3: the long-window rate O(T^{-1/α} + sqrt(log T / T)) is useful; a short remark on how α relates to typical retail price lower tails would help applied readers.
  4. Table 1 and the abstract: “OR policies” is clear in the text but may confuse readers outside operations research; a one-line definition at first use in the abstract would help.
  5. Section 6.4: the definition of ρ(v) across three valuation regimes is dense; a small table of ρ(v) formulas by (v vs p0, pU) would improve readability.
  6. References and related work: the connection to El-Yaniv et al. (2001) one-way trading is well placed; a sentence contrasting the three price primitives (pL, p0, pU) with classical one-way trading bounds would sharpen novelty.

Circularity Check

0 steps flagged

No significant circularity: threshold characterizations and robust guarantees are derived from stated primitives; empirical surplus is out-of-sample after pre-window calibration.

full rationale

The paper's load-bearing claims are (i) optimal/robust purchase policies under three information regimes and (ii) competitive performance of calibrated OR policies on held-out Keepa windows. The stationary threshold is obtained from the finite-horizon optimal-stopping variational inequality under Assumption 1 (Poisson adjustments, known H), yielding the ODE b'(s) = −λ E[(b(s)−P)+] with b(s)=v−m(s)—a standard free-boundary derivation, not a fit renamed as theory. The Bayesian threshold and information-gap bounds follow from the belief-state DP and a learn-then-act comparison under stated regularity (Assumptions 2–3); G is defined as oracle minus Bayesian value and bounded, not fitted to equal a target. Robust CR* = ρ(v) and Reg* = R̄(v) are matching upper/lower bounds via flash-sale distributions and explicit randomized threshold CDFs—competitive-analysis constructions, not self-definitional. Empirically, λ̂, empirical H, Dirichlet priors, and price bounds are estimated only from pre-window history; normalized surplus is realized surplus over offline OPT on the held-out path, which is never used for calibration. No uniqueness theorem is imported from overlapping authors to force the form; citations (Calvo, El-Yaniv et al., Bai–Perron, etc.) supply modeling context or screening tools, not the central characterizations. The derivation chain is therefore self-contained against its stated inputs and external evaluation benchmark.

Axiom & Free-Parameter Ledger

7 free parameters · 7 axioms · 1 invented entities

The central claims rest on standard optimal-stopping and competitive-analysis mathematics plus domain modeling choices that turn real posted-price paths into tractable primitives. Free parameters are those estimated or fixed for the empirical menu (intensity, empirical distributions, Dirichlet prior, robust bounds, heuristic screens, valuation multipliers). Axioms are the Poisson-Calvo structure, known intensity under Bayesian learning, exogenous non-strategic prices, known valuation and deadline, and the flash-sale adversary used for robust optimality. No new physical entities are postulated; 'OR policies' and 'policy menu' are organizational labels.

free parameters (7)
  • Poisson intensity λ̂
    Estimated as empirical frequency of pre-window price-adjustment timestamps; enters every stationary and Bayesian threshold.
  • Empirical price-adjustment distribution Ĥ
    Empirical measure of pre-window prices used as H in the stationary ODE and as residual distribution in the trend-aware extension.
  • Dirichlet prior counts / price bins
    Discretized Bayesian predictive model; prior counts initialized from pre-window prices and updated online.
  • Robust price bounds (p̂_L, p̂_U)
    Constructed from screened pre-window extrema and window-start price; fully determine the randomized threshold distribution.
  • Heuristic selector cutoffs (α_χ=0.5, n_0=15, κ=1, max breaks=5)
    Fixed diagnostic thresholds that route instances among Stationary/Bayesian/Trend-aware/Robust; chosen by authors, not cross-validated.
  • Valuation multiplier range [1.01, 1.50]
    Consumer valuation drawn as multiple of window-start price; shapes surplus scale and terminal decisions.
  • LLM lookback / sample-size choices
    LLM-OR selects lookback horizon and max recent observations used for calibration; free design choice of the agent.
axioms (7)
  • domain assumption Price adjustments arrive as a homogeneous Poisson process; post-adjustment prices are i.i.d. from a stationary distribution H (Assumption 1 / Calvo-style stickiness).
    Load-bearing for the stationary HJB, threshold ODE, and Bayesian DP; invoked throughout Sections 4–5.
  • domain assumption Adjustment intensity λ is known to the Bayesian agent; only the price-adjustment distribution is uncertain.
    Separates timing from distributional learning; stated at the opening of Section 5.
  • domain assumption Posted prices are exogenous; the agent’s buy/wait action does not affect future prices or availability.
    Stated in Section 3; rules out strategic seller responses and stock-outs.
  • domain assumption Consumer valuation v and deadline T are known and fixed; payoff is (v−p)+ or zero.
    Common environment in Section 3; isolates pure timing.
  • domain assumption Monotone likelihood-ratio order on the latent parameter family (Assumption 2) and posterior regularity / lower-tail / W_∞-Lipschitz conditions (Assumption 3).
    Required for Bayesian comparative statics and the long-window information-gap rate.
  • domain assumption In the robust regime, only price bounds (p_L,p_U) and initial price p_0 are trusted; performance is measured by competitive ratio or minimax regret against adversarial paths, with flash-sale processes as the worst-case subclass.
    Section 6 modeling choice that delivers the matching CR and regret bounds.
  • standard math Standard optimal-stopping variational inequalities, ODE comparison, Yao’s principle, and Brascamp–Lieb variance bounds.
    Background tools used in the e-companion proofs.
invented entities (1)
  • OR policy menu (Stationary / Bayesian / Robust / Trend-aware) independent evidence
    purpose: Organizes the three information regimes into implementable stopping rules selectable by heuristic or LLM.
    Terminological packaging of the derived policies; not a new physical object, but the paper’s operational contribution.

pith-pipeline@v1.1.0-grok45 · 51769 in / 4220 out tokens · 43224 ms · 2026-07-11T14:48:13.974309+00:00 · methodology

0 comments
read the original abstract

Agentic AI is shifting online shopping from search toward delegated purchasing, where autonomous buying agents monitor markets and decide when to buy on a consumer's behalf. We study the design of such strategic buying agents, which must decide when to purchase within a finite shopping window, translating price observations, the remaining time horizon, and beliefs about future price changes into a purchase policy. We formulate this problem across three information regimes: stationary, Bayesian, and robust, and treat the resulting optimal policies as a policy menu for implementation. In the stationary regime, price adjustments follow a Poisson arrival process with a known post-adjustment price distribution; the optimal policy is a dynamic purchase-threshold rule, with the threshold governed by an ordinary differential equation. In the Bayesian regime, the adjustment intensity is known, but the price-adjustment distribution is uncertain; the optimal rule remains threshold-based, now depending on posterior beliefs, and we bound the value of knowing the true distribution. In the robust regime, the agent has only price bounds and seeks worst-case protection; randomized threshold policies achieve optimal competitive-ratio and minimax-regret guarantees. We evaluate the proposed policies on Amazon price histories from Keepa (367 items, 48,933 timestamped observations) and examine their integration into language-model buying agents. The stationary and Bayesian policies perform competitively on mean normalized consumer surplus despite their stylized assumptions, while the robust policy performs best at the distribution's 10th percentile. Results suggest language models are better suited to selecting among regimes and calibration samples than to making buy-or-wait decisions directly.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

186 extracted references · 9 canonical work pages · 1 internal anchor

  1. [1]

    ArXiv:2508.02630v3

    Allouah A, Besbes O, Figueroa JD, Kanoria Y, Kumar A (2025) What is your AI agent buying? E valuation, biases, model dependence, and emerging implications for agentic e-commerce. ArXiv:2508.02630v3

    arXiv 2025

  2. [2]

    Management Sci

    Assun c \ a o JL, Meyer RJ (1993) The rational effect of price promotions on sales and consumption. Management Sci. 39(5):517--535

    1993

  3. [3]

    Manufacturing & Service Oper

    Aviv Y, Pazgal A (2008) Optimal pricing of seasonal products in the presence of forward-looking consumers. Manufacturing & Service Oper. Management 10(3):339--359

    2008

  4. [5]

    Econometrica 47--78

    Bai J, Perron P (1998) Estimating and testing linear models with multiple structural changes. Econometrica 47--78

    1998

  5. [6]

    Bai J, Perron P (2003) Computation and analysis of multiple structural change models. J. Appl. Econom. 18(1):1--22

    2003

  6. [7]

    Management Sci

    Bakos Y (1997) Reducing buyer search costs: Implications for electronic marketplaces. Management Sci. 43(12):1676--1692

    1997

  7. [8]

    Ball MO, Queyranne M (2009) Toward robust revenue management: Competitive analysis of online booking. Oper. Res. 57(4):950--963

    2009

  8. [9]

    Bawa VS (1975) Optimal rules for ordering uncertain prospects. J. Financ. Econ. 2(1):95--121

    1975

  9. [10]

    Besbes O, Zeevi A (2009) Dynamic pricing without knowing the demand function: Risk bounds and near-optimal algorithms. Oper. Res. 57(6):1407--1420

    2009

  10. [11]

    Birge JR, Chen HK, Keskin NB (2025) Markdown policies for demand learning with forward-looking customers. Oper. Res. 73(5):2550--2566, ://dx.doi.org/10.1287/opre.2019.0402

    work page doi:10.1287/opre.2019.0402 2025

  11. [12]

    Apple Podcasts, ://podcasts.apple.com/ca/podcast/stripes-john-collison-on-how-agentic-commerce-will/id1056200096?i=1000768091736, accessed May 17, 2026

    Bloomberg Odd Lots (2026) Stripe's john collison on how agentic commerce will reshape the internet. Apple Podcasts, ://podcasts.apple.com/ca/podcast/stripes-john-collison-on-how-agentic-commerce-will/id1056200096?i=1000768091736, accessed May 17, 2026

    2026

  12. [13]

    Borodin A, El-Yaniv R (1998) Online Computation and Competitive Analysis (Cambridge: Cambridge University Press)

    1998

  13. [14]

    Management Sci

    Brynjolfsson E, Smith MD (2000) Frictionless commerce? A comparison of Internet and conventional retailers. Management Sci. 46(4):563--585

    2000

  14. [15]

    Management Sci

    Cachon GP, Swinney R (2009) Purchasing, pricing, and quick response in the presence of strategic consumers. Management Sci. 55(3):497--511

    2009

  15. [16]

    Calvo GA (1983) Staggered prices in a utility-maximizing framework. J. Monet. Econ. 12(3):383--398

    1983

  16. [17]

    ArXiv:2603.20972v1

    Cao S, Hu M (2026) A solicit-then-suggest model of agentic purchasing. ArXiv:2603.20972v1

    arXiv 2026

  17. [18]

    Cavallo A (2018) Scraped data and sticky prices. Rev. Econ. Stat. 100(1):105--119

    2018

  18. [19]

    Manufacturing & Service Oper

    Chen H, Hu M, Wu J (2023) Intertemporal price discrimination via randomized promotions. Manufacturing & Service Oper. Management 25(3):1176--1194, ://dx.doi.org/10.1287/msom.2023.1194

    work page doi:10.1287/msom.2023.1194 2023

  19. [20]

    Proceedings of the 25th International Conference on World Wide Web, 1339--1349 (ACM)

    Chen L, Mislove A, Wilson C (2016 a ) An empirical analysis of algorithmic pricing on Amazon marketplace. Proceedings of the 25th International Conference on World Wide Web, 1339--1349 (ACM)

    2016

  20. [21]

    Management Sci

    Chen Q, Jasin S, Duenyas I (2016 b ) Real-time dynamic pricing with minimal and flexible price adjustment. Management Sci. 62(8):2437--2455, ://dx.doi.org/10.1287/mnsc.2015.2238

    work page doi:10.1287/mnsc.2015.2238 2016

  21. [22]

    Surveys in Operations Research and Management Science 20(1):1--18

    den Boer AV (2015) Dynamic pricing and learning: Historical origins, current research, and new directions. Surveys in Operations Research and Management Science 20(1):1--18

    2015

  22. [23]

    Algorithmica 30(1):101--139

    El-Yaniv R, Fiat A, Karp RM, Turpin G (2001) Optimal search and one-way trading online algorithms. Algorithmica 30(1):101--139

    2001

  23. [24]

    Feng Z, Dawande M, Janakiraman G, Qi A (2024) Technical note---dynamic pricing and learning with discounting. Oper. Res. 72(2):481--492, ://dx.doi.org/10.1287/opre.2023.2477

    work page doi:10.1287/opre.2023.2477 2024

  24. [25]

    Golosov M, Lucas RE Jr (2007) Menu costs and Phillips curves. J. Polit. Econ. 115(2):171--199

    2007

  25. [26]

    Hadar J, Russell WR (1969) Rules for ordering uncertain prospects. Amer. Econ. Rev. 59(1):25--34

    1969

  26. [27]

    Management Sci

    Harrison JM, Keskin NB, Zeevi A (2012) Bayesian dynamic pricing policies: Learning and earning under a binary prior distribution. Management Sci. 58(3):570--586, ://dx.doi.org/10.1287/mnsc.1110.1426

    Pith/arXiv arXiv doi:10.1287/mnsc.1110.1426 2012

  27. [28]

    Properties of the LWR model with time delay

    He M, Jennings NR, Leung HF (2003) On agent-mediated electronic commerce. IEEE Trans. Knowl. Data Eng. 15(4):985--1003, ://dx.doi.org/10.1109/TKDE.2003.1209014

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1109/tkde.2003.1209014 2003

  28. [29]

    Econometrica 74(6):1637--1673

    Hendel I, Nevo A (2006) Measuring the implications of sales and consumer inventory behavior. Econometrica 74(6):1637--1673

    2006

  29. [30]

    Proceedings of the Twenty-Second Annual ACM Symposium on Theory of Computing, 352--358 (ACM)

    Karp RM, Vazirani UV, Vazirani VV (1990) An optimal algorithm for on-line bipartite matching. Proceedings of the Twenty-Second Annual ACM Symposium on Theory of Computing, 352--358 (ACM)

    1990

  30. [31]

    Autonomous Agents and Multi-Agent Systems 5(3):255--287

    Kephart JO, Greenwald AR (2002) Shopbot economics. Autonomous Agents and Multi-Agent Systems 5(3):255--287

    2002

  31. [32]

    Keskin NB, Zeevi A (2014) Dynamic pricing with an unknown demand model: Asymptotically optimal semi-myopic policies. Oper. Res. 62(5):1142--1167

    2014

  32. [33]

    Kumar A, Manshadi V, Tumu A (2026) How much should a conversational recommender system converse? Available at SSRN: https://ssrn.com/abstract=6481459

    2026

  33. [34]

    Lippman SA, McCall JJ (1976) The economics of job search: A survey. Econ. Inq. 14(3):347--368, part II

    1976

  34. [35]

    Management Sci

    Liu Q, van Ryzin G (2008) Strategic capacity rationing to induce early purchases. Management Sci. 54(6):1115--1131

    2008

  35. [36]

    Maes P, Guttman RH, Moukas AG (1999) Agents that buy and sell. Commun. ACM 42(3):81--87, ://dx.doi.org/10.1145/295685.295716

    work page doi:10.1145/295685.295716 1999

  36. [37]

    McCall JJ (1970) Economics of information and job search. Q. J. Econ. 84(1):113--126

    1970

  37. [38]

    McDonald R, Siegel D (1986) The value of waiting to invest. Q. J. Econ. 101(4):707--727

    1986

  38. [39]

    Mehta A, Saberi A, Vazirani U, Vazirani V (2007) AdWords and generalized online matching. J. ACM. 54(5):22--es

    2007

  39. [40]

    Manufacturing & Service Oper

    Mersereau AJ, Zhang D (2012) Markdown pricing with unknown fraction of strategic customers. Manufacturing & Service Oper. Management 14(3):355--370, ://dx.doi.org/10.1287/msom.1120.0376

    work page doi:10.1287/msom.1120.0376 2012

  40. [41]

    Milgrom PR (1981) Good news and bad news: Representation theorems and applications. Bell J. Econ. 12(2):380--391

    1981

  41. [42]

    Management Sci

    Moon K, Bimpikis K, Mendelson H (2017) Randomized markdowns and online monitoring. Management Sci. 64(3):1271--1290, ://dx.doi.org/10.1287/mnsc.2016.2661

    work page doi:10.1287/mnsc.2016.2661 2017

  42. [43]

    Nair H (2007) Intertemporal price discrimination with forward-looking consumers: Application to the US market for console video-games. Quant. Mark. Econ. 5(3):239--292

    2007

  43. [44]

    Nakamura E, Steinsson J (2008) Five facts about prices: A reevaluation of menu cost models. Q. J. Econ. 123(4):1415--1464

    2008

  44. [45]

    ://openai.com/index/buy-it-in-chatgpt/, accessed May 17, 2026

    OpenAI (2025 a ) Buy it in chatgpt: Instant checkout and the agentic commerce protocol. ://openai.com/index/buy-it-in-chatgpt/, accessed May 17, 2026

    2025

  45. [46]

    ://openai.com/index/introducing-chatgpt-agent/, accessed April 20, 2026

    OpenAI (2025 b ) Introducing ChatGPT agent: Bridging research and action. ://openai.com/index/introducing-chatgpt-agent/, accessed April 20, 2026

    2025

  46. [47]

    ://openai.com/index/introducing-operator/, accessed April 20, 2026

    OpenAI (2025 c ) Introducing operator. ://openai.com/index/introducing-operator/, accessed April 20, 2026

    2025

  47. [48]

    Peskir G, Shiryaev A (2006) Optimal Stopping and Free-Boundary Problems (Basel: Birkh \"a user)

    2006

  48. [49]

    Management Sci

    Rosenfield DB, Shapiro RD, Butler DA (1983) Optimal strategies for selling an asset. Management Sci. 29(9):1051--1061

    1983

  49. [50]

    Rothschild M (1974) Searching for the lowest price when the distribution of prices is unknown. J. Polit. Econ. 82(4):689--711

    1974

  50. [51]

    Springer Series in Statistics (New York: Springer)

    Shaked M, Shanthikumar JG (2007) Stochastic Orders. Springer Series in Statistics (New York: Springer)

    2007

  51. [52]

    ArXiv:2507.21502v1

    Simchi-Levi D, Mellou K, Menache I, Pathuri J (2025) Large language models for supply chain decisions. ArXiv:2507.21502v1

    Pith/arXiv arXiv 2025

  52. [53]

    ://stripe.com/blog/developing-an-open-standard-for-agentic-commerce, accessed May 17, 2026

    Stripe (2025 a ) Developing an open standard for agentic commerce. ://stripe.com/blog/developing-an-open-standard-for-agentic-commerce, accessed May 17, 2026

    2025

  53. [54]

    ://stripe.com/blog/agentic-commerce-suite, accessed May 17, 2026

    Stripe (2025 b ) Introducing the agentic commerce suite: A complete solution for selling on AI agents. ://stripe.com/blog/agentic-commerce-suite, accessed May 17, 2026

    2025

  54. [55]

    ://stripe.com/en-sg/resources/more/agentic-commerce, accessed May 17, 2026

    Stripe (2026 a ) Agentic commerce: How AI agents are changing the way businesses buy and sell. ://stripe.com/en-sg/resources/more/agentic-commerce, accessed May 17, 2026

    2026

  55. [56]

    ://stripe.com/blog/giving-agents-the-ability-to-pay, accessed May 17, 2026

    Stripe (2026 b ) Giving agents the ability to pay. ://stripe.com/blog/giving-agents-the-ability-to-pay, accessed May 17, 2026

    2026

  56. [57]

    Management Sci

    Su X (2007) Intertemporal pricing with strategic customer behavior. Management Sci. 53(5):726--741

    2007

  57. [58]

    Gronau N, Krasnova H, Pousttchi K, Heine M, eds., Entwicklungen, Chancen und Herausforderungen der Digitalisierung: Band 1, Proceedings der 15

    Weber M, Kowalkiewicz M, Weking J, B \"o hm M, Krcmar H (2020) When algorithms go shopping: Analyzing business models for highly autonomous consumer buying agents. Gronau N, Krasnova H, Pousttchi K, Heine M, eds., Entwicklungen, Chancen und Herausforderungen der Digitalisierung: Band 1, Proceedings der 15. Internationalen Tagung Wirtschaftsinformatik 2020...

    2020

  58. [59]

    Econometrica 90(2):831--858

    Williams KR (2022) The welfare effects of dynamic pricing: Evidence from airline markets. Econometrica 90(2):831--858

    2022

  59. [60]

    Yao S, Chen H, Yang J, Narasimhan K (2022) WebShop : Towards scalable real-world web interaction with grounded language agents. Adv. Neural Inf. Process. Syst., volume 35, 20744--20757

    2022

  60. [61]

    International Conference on Learning Representations

    Yao S, Zhao J, Yu D, Du N, Shafran I, Narasimhan K, Cao Y (2023) ReAct : Synergizing reasoning and acting in language models. International Conference on Learning Representations

    2023

  61. [62]

    Manufacturing & Service Oper

    Zhang H, Jasin S (2022) Online learning and optimization of (some) cyclic pricing policies in the presence of patient customers. Manufacturing & Service Oper. Management 24(2):1165--1182, ://dx.doi.org/10.1287/msom.2021.0979

    work page doi:10.1287/msom.2021.0979 2022

  62. [63]

    2024 , publisher=

    Price Delegation with Learning Agents , author=. 2024 , publisher=

    2024

  63. [64]

    2021 , publisher=

    Learning-Based Robust Optimization: Procedures and Statistical Guarantees , author=. 2021 , publisher=

    2021

  64. [65]

    2022 , publisher=

    Data Pooling in Stochastic Optimization , author=. 2022 , publisher=

    2022

  65. [66]

    2021 , publisher=

    Small-Data, Large-Scale Linear Optimization with Uncertain Objectives , author=. 2021 , publisher=

    2021

  66. [67]

    2009 , publisher=

    Toward Robust Revenue Management: Competitive Analysis of Online Booking , author=. 2009 , publisher=

    2009

  67. [68]

    Optimal entropy-transport problems and a new Hellinger--Kantorovich distance between positive measures , author=. Invent. Math. , volume=. 2018 , publisher=

    2018

  68. [69]

    Free boundaries in optimal transport and Monge-Ampere obstacle problems , author=. Ann. Math. , pages=. 2010 , publisher=

    2010

  69. [70]

    The optimal partial transport problem , author=. Arch. Ration. Mech. Anal. , volume=. 2010 , publisher=

    2010

  70. [71]

    Existence, uniqueness and efficiency of equilibrium in hedonic markets with multidimensional types , author=. Econ. Theory , volume=. 2010 , publisher=

    2010

  71. [72]

    2008 , publisher=

    Optimal transport: old and new , author=. 2008 , publisher=

    2008

  72. [73]

    2024 , publisher=

    Demand Equilibria in Spatial Service Systems , author=. 2024 , publisher=

    2024

  73. [74]

    Extremal dependence concepts , author=

  74. [75]

    Inequalities for distributions with given marginals , author=. Ann. Probab. , pages=. 1980 , publisher=

    1980

  75. [76]

    2016 , publisher=

    Optimal transport methods in economics , author=. 2016 , publisher=

    2016

  76. [77]

    arXiv preprint arXiv:1808.01962 , year=

    Semi-discrete unbalanced optimal transport and quantization , author=. arXiv preprint arXiv:1808.01962 , year=

    arXiv

  77. [78]

    1977 , publisher=

    Plemmons, Robert J , journal=LAA, volume=. 1977 , publisher=

    1977

  78. [79]

    Monge, Gaspard , journal=. M

  79. [80]

    On the translocation of masses , author=. Dokl. Akad. Nauk. USSR (NS) , volume=

  80. [81]

    2024 Winter Simulation Conference (WSC) , pages=

    Introduction to Optimal Transport , author=. 2024 Winter Simulation Conference (WSC) , pages=. 2024 , organization=

    2024

Showing first 80 references.