Skip to content
ScoAI: OR-Native Decision Intelligence for Supply Chains

ScoAI: OR-Native Decision Intelligence for Supply Chains

This article reviews the ScoAI framework, which integrates supply chain context, operations research, and AI to create verifiable, executable decisions by embedding decision structure into model interfaces, decomposing decision errors, and emphasizing execution feedback over prediction accuracy.

JD Cloud Developers
JD Cloud Developers
JD Cloud Developers

Background

Supply chain inventory management requires decisions on replenishment quantities, inventory allocation, and execution timing under demand and supply uncertainty. Foundation models and agents can estimate demand, parse supplier information, and generate candidate plans, but model outputs are not equivalent to business decisions. Only when embedded in explicit business objectives, warehouse capacity and carton constraints, transportation conditions, and execution processes can they form implementable, verifiable operational plans.

The paper ScoAI: Toward OR-Native Decision Intelligence addresses the gap between model outputs and actual decisions. It proposes ScoAI = SC + OR + AI : supply chain business defines the decision context, available resources, and action consequences; operations research transforms objectives, constraints, and cross-stage coordination into computable problems; AI provides reusable estimates and judgments across tasks. The three are not independent capability additions but jointly constitute a computational loop centered on business decision execution, verification, and update. As a Perspective, the paper focuses on a methodological question: when foundation models and agents are applied to inventory, allocation, and planning, how to move from generating superficially plausible answers to supporting verifiable, executable, and traceable decisions.

Core Idea

The paper's main contribution is not a new network architecture but elevating decision structure to be part of the model interface. In common pipelines, models output predictions while objectives, constraints, and execution rules are handled externally by business systems. ScoAI advocates making these elements and their relationships explicit from the start of modeling, so they can be independently verified and persist through training, solving, and runtime updates.

This requirement is summarized as OR-native (运筹原生) . OR-native does not merely mean plugging in a solver; it requires business objectives, optional actions, uncertain factors, information available at decision time, and execution feedback to enter the model or agent interface in explicit form. Thus model outputs can directly serve subsequent computation, and when business rules change, the system can identify exactly which components need adjustment.

This distinction is critical for inventory systems. Inventory conservation is a deterministic business relationship that need not be learned; warehouse capacity, carton rules, and service-level requirements are current operational rules that must be explicitly expressed and version-controlled; demand, lead times, substitution relationships, and supplier responses are uncertain and require data-driven estimation. The paper terms the combination of learning components with explicit operational mechanisms data–mechanism fusion (数据—机制融合) .

Data–mechanism fusion is not simply appending an optimizer after a prediction model. The structure of business loss determines whether the prediction stage should focus on mean, quantile, or tail risk; solver-identified infeasibility can correct modeling assumptions or candidate plans; actual shipment and receipt records should update the system state for the next decision round. Different feedback acts on different objects and cannot be lumped together as model error.

The paper further decomposes decision error into three categories: (1) high-quality plans missing from the candidate set, (2) model failure to correctly compare candidates' business value, and (3) insufficient search within limited compute budget. Each error type demands a different improvement path. Improving solver capability only enhances search quality; it cannot add missing candidates or fix erroneous value estimation.

Cross-stage decisions require propagating downstream impacts upstream. Assortment decisions alter subsequent replenishment demand; replenishment plans consume transportation and receiving capacity; thus a locally low-cost choice may increase total chain cost. The paper introduces completion value to measure a local plan's potential to form a high-quality complete plan, and uses resource prices to pass the opportunity cost of shared capacity to each stage. For businesses with full-carton requirements, integer variables, and shared resources, the final plan must still pass joint feasibility checks.

System Design

The proposed agent architecture does not require the foundation model to handle the entire decision flow. Models are suited for understanding text, estimating demand and lead times, and proposing candidate plans; deterministic date calculations, inventory conservation, full-carton constraints, warehouse capacity limits, and shared-resource coordination remain the responsibility of programs and solvers; permission checks, resource reservation, and business submission continue to be managed by business systems.

Structured judgment models can be integrated via well-defined interfaces: they read authorized state information, judge well-defined problems, and submit option probabilities or rankings to the planner. Model confidence alone does not constitute execution permission. The final plan must still pass verification across five layers: semantic, informational, value, computational, and operational.

Execution records are an indispensable part of this architecture. The system should separately store the plan proposed by the model or planner, the approval result, and the actual executed action, and update business state with the actual action rather than the original plan. For inventory management, this affects the accuracy of on-hand and in-transit quantities for the next period and determines which subsequent observations can support analysis and judgment.

Supply Chain Scenarios

The paper draws on existing JD.com research to show that the described interfaces already have technical foundations in various supply chain scenarios. In inventory control, dynamic programming decisions have been used as supervision signals to train replenishment policies adapted to actual operating costs; regional and forward warehouse assortment allocation incorporates demand forecasting, shared inventory, warehouse capacity, and fulfillment cost into a single optimization problem; generative planning connects business requests, data, and analytical tools; ORLA involves solvers in optimization model generation and selection; SCOPE coordinates assortment, sourcing, replenishment cycles, and fulfillment paths across the full chain.

These works demonstrate multiple connection modes between learning and optimization: OR solutions as training supervision, operating costs as learning signals, solver feedback correcting model outputs, downstream results evaluating upstream choices. The paper notes these works address different scenarios with different experimental designs, so their results cannot be aggregated into an overall ScoAI business benefit; they are cited to illustrate how different interface types are realized in concrete tasks.

This methodology aligns with two recent studies from the same team. One recasts the destroy step of large neighborhood search as a learnable structured problem [2]; the other explores how language models can provide prior knowledge and policy structure at the design stage of lost-sales inventory control, while an OR controller handles actual execution [3]. Though their tasks and methods differ, both embody the principle that AI should provide reusable capabilities in well-defined computational roles, not replace existing constraint expression, optimization solving, and execution control mechanisms.

A constructed example in the paper shows that business rule changes cannot be handled solely by the model's semantic understanding. When a store shipment limit changes from 4 to 2, the old plan may still receive a high model score but is no longer in the new feasible region. Demand estimation and inventory conservation remain unchanged; the allowed action space has changed. The correct response is to update the constraint and re-solve, not to prioritize retraining the prediction model.

Full-carton constraints illustrate the same issue. Rounding a per-unit shipment plan may simultaneously leave store demand unmet and violate total inventory constraints. The more reliable approach is to write carton rules directly into the joint decision model and re-compute. Such basic requirements in warehousing and distribution indicate that explicit decision structures cannot be replaced by free-text generation.

Business Implications

ScoAI's significance lies not in replacing existing business systems with general models, but in clarifying the applicable scope of AI capabilities within the decision chain. Demand and lead times can be estimated from data; inventory conservation and carton constraints must be explicitly enforced; supplier text can be parsed by models; whether to expedite still requires comprehensive cost and resource evaluation; candidate plans can be generated by models, but before submission they must pass verification, resource reservation, and permission audit. For existing inventory systems, this path is more implementable than directly replacing mature algorithms.

The paper provides a targeted diagnostic framework. When policy performance degrades, one can separately check: (1) whether the candidate set contains high-quality plans, (2) whether candidate value ranking is accurate, and (3) whether the search process was sufficient. Warehouse capacity changes require constraint updates; lead-time distribution changes require response model updates; discrepancies between actual and planned shipments require execution record corrections. Many issues vaguely attributed to "model drift" may actually occur at different stages of the decision chain. Accordingly, data recording must cover the complete decision process. Saving only inputs and final results is insufficient for accurate attribution; one must also link the rule version used at decision time, candidate plans, system recommendations, approval results, actual execution, and subsequent observations to distinguish the impacts of model, solver, human intervention, and execution deviation. System evaluation should not be limited to prediction accuracy and model invocation cost, but should also assess decision quality, hard constraint satisfaction, task coverage, and end-to-end total cost.

General models are not suitable for all supply chain tasks. For replenishment scenarios with stable rules and clear problem boundaries, mature specialized OR processes often remain more reliable and economical. Reusable models show advantage only when the same understanding, estimation, or modeling capability can serve multiple related decisions; their value must still be judged by decision effectiveness, task coverage, and overall cost.

Summary

ScoAI focuses not on whether AI can enter supply chains, but on what should be learned from data, what relationships must be computed explicitly, and what results need verification against execution records after AI participates in supply chain decisions. The basic division of labor is: supply chain business defines the problem, resources, and action consequences; operations research expresses objectives, constraints, and coordination; AI provides reusable estimation and judgment. What truly improves subsequent decisions is not only the model's initial output but also the actual actions taken and the business feedback they generate.

Paper and Related Works

[1] ScoAI: Toward OR-Native Decision Intelligence<br>Authors: Yang Liu, Yulin Huang, Xue Yu, Haowei Zhou, Jiaxin Wang, Qiqi Yan, Tianqi Zhou, Yi Yao, Zhongxuan Han, Shaomin Li, Jianshen Zhang, Zhe Chen, Yongzhi Qi*, Wei Qi, Zuo-Jun (Max) Shen.

[2] Recasting the Destroy Step of Large Neighborhood Search as Dense Segmentation<br>Authors: Yang Liu, Yulin Huang, Jianshen Zhang, Yongzhi Qi*.

[3] SabreAgent: Language Models at Design Time for Lost-Sales Inventory Control<br>Authors: Yang Liu, Yulin Huang, Xue Yu, Jiong Dong, Jianshen Zhang, Yongzhi Qi*.

Original Source

Signed-in readers can open the original source through BestHub's protected redirect.

Sign in to view source
Republication Notice

This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactadmin@besthub.devand we will review it promptly.

supply chainOperations ResearchAgentsDecision Intelligencefoundation modelsData-Mechanism FusionOR-NativeScoAI
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.