AI in Supply Chain Management Training Course

Participants can map priority supply chain decisions, inputs, outputs and accountable roles.

• SCOR processes connected to planning, sourcing, making, delivering and returning
• Decision inventory created for tactical, operational and strategic supply chain choices
• Inputs, outputs and decision owners documented using a SIPOC map
• Exception workflows mapped for shortages, late deliveries and demand shocks
• Planning horizons aligned with data refresh cycles and intervention windows
• Decision rights separated between planners, managers, systems and suppliers

Participants can identify and prioritise AI use cases using value, feasibility and risk criteria.

• Use cases generated from costly decisions, bottlenecks and recurring exceptions
• Value-versus-feasibility matrix scores benefits, data availability and implementation complexity
• Problem statements specify users, decisions, actions and expected business outcomes
• Baseline performance captured before proposing AI-enabled process changes
• Risk scoring considers safety, service, financial and reputational consequences
• Use-case portfolio balanced between quick wins and strategic capabilities

Participants can assess data readiness for an AI supply chain use case.

• Master data domains identified for products, suppliers, locations and customers
• Data lineage traced from source transaction through planning and reporting layers
• Completeness, accuracy, timeliness and consistency measured using data quality rules
• Missing values, duplicates and outliers assessed before model development
• Data ownership assigned across supply chain, finance, technology and vendors
• ISO 8000 principles applied to product and party master-data quality

Participants can specify data preparation steps and useful supply chain model features.

• Time-series data structured by product, location, customer and planning period
• Promotions, holidays and stockouts flagged before demand-model training
• Lead-time variability calculated from purchase orders, receipts and supplier confirmations
• Feature engineering combines lagged demand, seasonality and external demand drivers
• Train, validation and test datasets separated without future-data leakage
• Data dictionaries document definitions, transformations, units and business assumptions

Participants can evaluate machine-learning forecasting approaches against planning requirements.

• Statistical baselines compared with regression, tree-based and neural forecasting models
• Forecast granularity selected by product, location, channel and planning horizon
• MAPE, WAPE, bias and RMSE interpreted for demand-planning decisions
• Intermittent demand handled using Croston-style methods and segmentation rules
• Forecast value added separates model contribution from planner judgement
• Demand sensing inputs assessed for availability, timeliness and causal relevance

Participants can connect AI predictions to inventory policy and replenishment decisions.

• Safety stock recalculated using demand variability, lead-time variability and service targets
• Reorder points linked to probabilistic demand forecasts and replenishment lead times
• Multi-echelon inventory concepts applied to central, regional and local stocking points
• ABC-XYZ segmentation determines suitable inventory policies and model attention
• Stockout, obsolescence and working-capital trade-offs quantified through scenario analysis
• Recommended orders checked against constraints, minimums and supplier capacity

Participants can apply AI methods to spend analysis, supplier assessment and sourcing decisions.

• Spend classification uses supervised learning, taxonomy rules and confidence thresholds
• Supplier records matched through entity resolution and duplicate-detection techniques
• Natural language processing extracts obligations from contracts and supplier documents
• Supplier risk indicators combine delivery, quality, financial and external-event signals
• Should-cost inputs analysed using commodity indices, specifications and purchase history
• Procurement recommendations reviewed for fairness, auditability and approval authority

Participants can identify AI applications for warehouse capacity, labour and fulfilment performance.

• Order profiles segmented by velocity, line count, cube and handling requirements
• Labour demand forecasts built from inbound, picking and shipping workload drivers
• Slotting recommendations consider travel distance, compatibility and replenishment frequency
• Computer vision use cases assessed for counting, damage and safety inspection
• Queue analytics identifies congestion at receiving, packing and dispatch points
• Warehouse model outputs tested against safety, ergonomic and operational constraints

Participants can assess AI-enabled transport planning and disruption-response applications.

• Estimated time of arrival models use route, carrier and event data
• Vehicle routing optimisation balances distance, capacity, delivery windows and emissions
• Carrier performance scorecards combine on-time delivery, damage and tender acceptance
• Exception alerts prioritised by customer impact, recovery options and financial exposure
• Digital twin concepts tested for network capacity and disruption scenarios
• Control-tower workflows define escalation, intervention and closure responsibilities

Participants can evaluate model performance, reliability and operational suitability before deployment.

• Classification metrics include precision, recall, F1 score and confusion matrices
• Regression metrics compared against baselines, tolerance bands and business cost
• Backtesting validates forecasts using historical periods and realistic information availability
• Model calibration checked where probability scores trigger operational actions
• Error analysis identifies products, locations and conditions with weak performance
• Acceptance criteria combine technical accuracy, user trust and decision value

Participants can design explainable AI outputs and appropriate human oversight controls.

• Explainability requirements vary by decision consequence, user role and regulation
• SHAP values interpreted to explain individual predictions and feature influence
• Confidence scores displayed alongside recommendations, assumptions and source data
• Human-in-the-loop checkpoints set for overrides, approvals and high-risk exceptions
• Override reasons captured to improve models and strengthen audit trails
• User interfaces tested for clarity, actionability and cognitive workload

Participants can identify and control responsible AI risks in supply chain applications.

• NIST AI RMF functions applied to supply chain AI controls
• Bias testing examines supplier, geography, customer and product-group outcomes
• Privacy obligations assessed for employee, customer and supplier information
• Cybersecurity threats include data poisoning, prompt injection and model theft
• Model risk registers record assumptions, controls, owners and review dates
• Incident response procedures define containment, investigation and stakeholder communication

Participants can define roles, supplier requirements and governance for AI delivery.

• Product owner, data owner and model owner accountabilities clearly assigned
• Cross-functional governance forums review value, risk, performance and change requests
• Vendor due diligence assesses security, data rights, support and model transparency
• Statements of work define acceptance tests, documentation and service levels
• Build, buy and partner options compared against capability and total cost
• Change-control procedures manage model updates, data changes and retraining triggers

Participants can create a phased implementation plan with measurable benefits and controls.

• Business cases quantify service, cost, inventory, productivity and risk benefits
• Benefit hypotheses linked to baseline measures, owners and measurement periods
• Pilot scope limits products, locations, users and decision types
• Implementation roadmaps sequence data work, model development, testing and adoption
• Stage gates require evidence before expansion, automation or further investment
• Adoption plans address training, communications, resistance and planner confidence

Participants can apply generative AI safely to supply chain documentation and analysis tasks.

• Generative AI use cases include document summarisation, query support and drafting
• Prompt structures specify role, task, context, constraints and output format
• Retrieval-augmented generation connects responses to approved supply chain knowledge sources
• Hallucination controls require source checking, validation and restricted use cases
• Confidential information protected through access controls and approved AI environments
• Generated outputs reviewed against policy, technical accuracy and commercial sensitivity

Participants can plan stakeholder adoption for AI-supported process and role changes.

• Stakeholder maps identify affected planners, buyers, operators, leaders and partners
• Process impacts assessed across tasks, handoffs, skills and decision authority
• Training plans distinguish system use, judgement skills and escalation procedures
• Communications explain decision rationale, expected benefits and user responsibilities
• Adoption metrics track usage, overrides, adherence and realised operational outcomes
• Feedback loops convert user observations into prioritised improvement requirements

Participants can build, review and present a complete governed AI supply chain initiative proposal.

• Priority decision and use case defined with measurable operational objectives
• Data readiness assessment documents sources, quality gaps, ownership and access
• Model approach specifies target, features, metrics, baselines and acceptance criteria
• Process design maps recommendations, user actions, overrides and escalation routes
• Risk controls cover bias, privacy, cybersecurity, monitoring and incident response
• Implementation roadmap presents pilot scope, costs, benefits, milestones and governance