BCGProcess redesign required
AI alone is not enough
BCG’s 2026 supply chain planning research finds few companies have converted planning-system and AI investment into consistent performance gains without redesigning process and behaviour.
Program 11 · Domain specialisation
SCIKIQ Certified Data and AI EngineerManufacturing & Supply Chain
Manufacturers have invested heavily in planning systems and seen little from AI on top of them. The reason is consistent: AI sat in a reporting layer while decisions stayed in ERP. The shift that works is moving AI into the execution layer, where it can validate constraints and trigger real workflow.
- 80 hrstaught
- 40hands-on AI labs
- 8assessment tests
- +20 hrs2 E2E capstones
15-45%Category cost reduction reported from AI in procurementBCG
25-40%Procurement productivity liftMcKinsey
60+Purpose-built supply chain agents coordinated by SAP Joule assistantsSAP
2026Autonomous supply chain management reaching phased general availabilitySAP
01
Part 1 · The business
The value chain, the people, the trends
Plan, source, make, deliver — and the two disciplines that cut across all of them, quality and service. Part 1 maps where decisions are made and why AI has struggled to reach them.
The manufacturing & supply chain value chain
Six stages from planning to service. Select one.
01
Plan
Demand planning, supply planning, S&OP and inventory strategy.
Decisions made here
- What will be demanded?
- What do we build and hold?
- How do we respond to a constraint?
Where it breaks today
- Planning systems that optimise on stale assumptions
- Planner overrides that are never captured
- Trade-offs made sequentially rather than jointly
What is measured
- Forecast accuracy
- Service level
- Inventory turns
Where AI landsMulti-signal forecasting, exception-based planning and joint trade-off optimisation.
Forecast accuracyService levelInventory turns
02
Source
Procurement, supplier management, contracts and category strategy.
Decisions made here
- Who do we buy from?
- At what price and terms?
- What is our supply risk?
Where it breaks today
- Spend data too fragmented to analyse
- Contract terms unknown at the point of buying
- Supplier risk discovered after disruption
What is measured
- Category cost
- Supplier risk
- Contract compliance
Where AI landsSpend classification, contract term extraction, should-cost analysis and supplier risk monitoring.
Category costSupplier riskContract compliance
03
Make
Production execution, OEE, changeovers, maintenance and shop-floor operations.
Decisions made here
- What do we run next?
- Why did we lose that time?
- Is this equipment healthy?
Where it breaks today
- Downtime causes recorded inconsistently
- Scheduling that ignores real constraints
- Shift knowledge that does not transfer
What is measured
- OEE
- First-pass yield
- Scrap rate
- Schedule adherence
Where AI landsDowntime cause analysis, schedule optimisation, predictive maintenance and shift handover support.
OEEFirst-pass yieldScrap rateSchedule adherence
04
Quality
Inspection, non-conformance, CAPA, supplier quality and continuous improvement.
Decisions made here
- Is this conforming?
- What caused the defect?
- Do we need to contain?
Where it breaks today
- Defect root cause taking weeks
- Quality data disconnected from process data
- Containment decisions made late
What is measured
- First-pass yield
- Cost of poor quality
- Containment time
Where AI landsVisual inspection, root-cause support correlating process and quality data, and CAPA drafting.
First-pass yieldCost of poor qualityContainment time
05
Deliver
Warehousing, transport, logistics execution and order fulfilment.
Decisions made here
- How does this order ship?
- What is at risk of missing?
- How do we recover?
Where it breaks today
- Exceptions discovered by the customer
- Transport decisions optimised in isolation
- Manual expediting at volume
What is measured
- On-time in-full
- Logistics cost
- Expedite spend
Where AI landsException detection, dynamic routing support and proactive customer communication.
On-time in-fullLogistics costExpedite spend
06
Service & aftermarket
Field service, spare parts, warranty and installed-base management.
Decisions made here
- What does this asset need?
- Is this warranty valid?
- What parts do we stock where?
Where it breaks today
- Service history in unstructured notes
- Parts demand hard to forecast
- Warranty claims processed manually
What is measured
- First-time fix
- Parts availability
- Warranty cost
Where AI landsService history mining, parts demand forecasting and warranty claim triage.
First-time fixParts availabilityWarranty cost
Who you are building for
The plant manager and the planner decide whether this becomes real. Select one to light the stages they own.
What the strategy houses are seeing
A sector where the technology is ready and the operating model usually is not.
BCG15-45% category cost
Procurement is the clearest value pool
AI in procurement is associated with 15-45% category cost reduction, with McKinsey reporting 25-40% productivity lift in the same function.
SAPExecution layer
AI moves from reporting to execution
The practical shift is AI operating inside transactional workflows — validating constraints and triggering actions — rather than describing what happened.
SAP60+ agents coordinated
Agent fleets, not single agents
Autonomous supply chain management coordinates dozens of purpose-built agents across planning, manufacturing, logistics and asset management.
BCGCompetitiveness, not efficiency
The factory of the future changes the economics
AI applied from procurement through demand planning and process engineering is reshaping manufacturing competitiveness, not just plant efficiency.
McKinseyJoint, not sequential
Agents break sequential trade-offs
Agents identify solutions that sequential human workflows cannot reach, because they can hold more constraints at once than a handoff chain can.
Where these numbers come from — 4 sources
Consulted when this program was written, in July 2026. Each entry names the claim on this page that it backs, so the pairing stays checkable if the copy is later edited.
- BCGSupply Chain Planning 2026: Why AI Alone Isn’t EnoughProcess redesign required to convert AI investment
- BCGHow the Factory of the Future Is Reshaping Manufacturing CompetitivenessAI reshaping competitiveness, not just plant efficiency
- IndustrySAP at Hannover Messe 2026: Supply Chain AI AgentsAI moving from reporting layer to execution layer
- IndustryAI Procurement Cuts Material Cost 15-45%15-45% category cost reduction in procurement
02
Part 2 · Data and meaning
Domains, ontology, taxonomy and governed terms
Manufacturing architecture spans two worlds: ERP and planning systems on one side, shop-floor systems on the other. The engineer’s job is usually to make them agree on the same reality.
The data landscape
Six families across the ERP/shop-floor divide.
ERP & planning
Orders, inventory, BOMs, routings and the planning engine outputs.
SAP S/4HANA, IBP, APSMES & shop floor
Production execution, downtime events, quality checks and machine states.
MES, SCADA, historiansProcurement & contracts
Spend, suppliers, contracts, catalogues and risk data.
Ariba, Coupa, CLMLogistics & warehouse
Shipments, warehouse movements, carrier data and tracking.
WMS, TMS, carrier APIsQuality & maintenance
Inspection results, non-conformances, work orders and asset history.
QMS, EAM, SAP PMExternal signals
Supplier risk, logistics disruption, commodity prices and demand signals.
Risk feeds, market dataThe semantic layer
ERP, MES and WMS each believe they own material identity. Reconciling them into one graph — material, location, order, equipment — is usually the critical path of an industrial deployment, and the reason recommendations are trusted or ignored.
Data domains
Six subject areas, each with the entities it holds, its critical data elements and the function accountable for it.
Material
Everything bought, made, held or sold, and how items compose.
MaterialBill of materialsRoutingBatch/lot
- Critical data elements
- Material number; BOM component quantity; Lot number
- Owned by
- Master data
Location & network
Plants, warehouses, suppliers, customers and the lanes between them.
PlantStorage locationSupplierCustomerLane
- Critical data elements
- Plant code; Supplier identifier; Lead time
- Owned by
- Supply chain
Demand & supply
Requirements, plans, orders and the commitments they create.
ForecastPlanned orderPurchase orderSales order
- Critical data elements
- Requirement date; Order quantity; Confirmed date
- Owned by
- Planning
Production
Execution on the shop floor and what it consumed and produced.
Production orderOperationWork centreDowntime event
- Critical data elements
- Order number; Yield; Downtime reason code
- Owned by
- Manufacturing
Quality
Inspection, non-conformance and corrective action.
Inspection lotDefectNon-conformanceCAPA
- Critical data elements
- Defect code; Disposition; Root cause category
- Owned by
- Quality
Asset & maintenance
The equipment that makes it and the work done to keep it running.
EquipmentFunctional locationMaintenance orderFailure mode
- Critical data elements
- Equipment number; Criticality; Failure code
- Owned by
- Maintenance
Ontology & knowledge graph
The entities and the typed relationships between them — what the agent traverses instead of guessing joins. Select any entity.
Core entityEventReference / classification
Core entity
Material
Any bought, made or sold item.
- Material composed via Bill of materials
- Bill of materials consumes Material
- Supplier supplies Material
- Purchase order delivers Material
- Forecast predicts demand for Material
- Production order produces Material
- Sales order requires Material
Reference / classification
Bill of materials
The component structure of a material.
- Bill of materials consumes Material
- Material composed via Bill of materials
Core entity
Plant
A manufacturing or storage site.
- Production order runs at Plant
Core entity
Supplier
A source of purchased material.
- Supplier supplies Material
- Purchase order ordered from Supplier
Core entity
Customer
The recipient of finished goods.
- Sales order placed by Customer
Event
Forecast
Expected demand for a material at a location.
- Forecast predicts demand for Material
Event
Purchase order
A commitment to buy from a supplier.
- Purchase order ordered from Supplier
- Purchase order delivers Material
Event
Production order
An instruction to manufacture.
- Production order produces Material
- Production order runs at Plant
- Production order uses Equipment
- Production order yields Defect
- Downtime event delays Production order
Event
Sales order
A customer commitment to buy.
- Sales order placed by Customer
- Sales order requires Material
Core entity
Equipment
A machine used in production.
- Equipment suffers Downtime event
- Production order uses Equipment
Event
Downtime event
A period of lost production with a reason.
- Downtime event delays Production order
- Equipment suffers Downtime event
Event
Defect
A quality non-conformance found on output.
- Production order yields Defect
Taxonomies
The classification hierarchies that make records comparable across systems.
Material classification
- Material type
- Group
- Class
- Material
Raw → Metals → Stainless sheet → SS304 2mm
Downtime reason
- Category
- Reason
- Sub-reason
Unplanned → Equipment failure → Servo drive fault
Spend category
- Level 1
- Level 2
- Level 3
Indirect → MRO → Bearings and seals
Governed business terms
The definitions an agent must use rather than invent. Most wrong answers in this industry are a term used loosely.
- Material vs part number
- A material is your internal master record; a supplier part number is theirs. Mapping them is the integration.
- OEE
- Availability x performance x quality. Each factor must be defined identically across sites or the number is meaningless.
- First-pass yield
- Units produced correctly without rework, as a proportion of units started.
- OTIF
- On-time in-full — delivery measured against the customer-confirmed date, not the internal plan.
- Planned vs confirmed date
- What planning intends versus what supply has committed. Recommendations must use confirmed.
- Functional location
- The position in the plant hierarchy, persisting when the equipment installed there changes.
- Safety stock
- Inventory held to absorb variability, distinct from cycle stock covering expected demand.
- Should-cost
- A modelled estimate of what a purchased item ought to cost given inputs and process.
03
Part 2 · The architecture
How it is actually built
The layers a deployment needs, what each one holds, and what the engineer owns there.
The reference architecture
Seven layers, joining planning systems to shop-floor reality.
01/07
Sources & systems
ERP, MES, WMS, EAM and procurement systems, each with different latency and access models.
Takes in
Nothing. This is the upstream edge.
Hands on
Order, execution and movement feeds tied to one material and asset hierarchy.
SystemWhat it doesKey business entities
ERP
Materials, bills of materials, orders and costs. The commercial truth of what was planned, ordered and paid for.
MaterialBOMOrderCost
Manufacturing execution
What actually happened on the line: work orders, operations, genealogy and quality results, at a resolution the ERP never sees.
Work orderOperationGenealogyDefect
Planning systems
Demand, supply and constrained plans. The place a constraint is either visible on planning day or discovered on delivery day.
ForecastPlanConstraintAllocation
Warehouse and transport
Stock movements, shipments, carriers and events. Where in-transit visibility either exists or is quietly assumed.
ShipmentMovementCarrier event
Quality management
Inspections, non-conformances, CAPA and supplier quality. Links a defect back to a lot and a lot back to a supplier.
InspectionNon-conformanceLotSupplier
Plant floor and IIoT
Machine telemetry, historians and downtime capture. High-volume, low-context, and meaningless without the asset hierarchy.
MachineTelemetryDowntime event
What the engineer owns here
Getting shop-floor and ERP data onto a shared time base and product identity.
Where this layer fails
Trusting the ERP for what happened on the line. It records what was planned and settled; the execution system records reality, and the two are never identical.
02/07
Ingestion & harmonisation
Streaming shop-floor events alongside batch ERP extracts, harmonised to one model.
Takes in
Read-mostly access paths, change feeds and document streams from the systems of record.
Hands on
Replayable, resolved, quality-checked records with their lineage back to the source row.
SystemWhat it doesKey business entities
Change data capture
Reads the source’s own change log rather than polling it, so the platform sees every state a record passed through instead of only where it ended up.
Change eventOperation typeCommit timestamp
Landing and raw zone
Every extract kept as it arrived, immutable and timestamped. The thing you replay from when a downstream definition turns out to have been wrong.
Raw extractBatchWatermark
Document processing
Layout-aware extraction over the unstructured half of the estate: sectioning, tables, signatures, and the page reference every later citation depends on.
DocumentSectionPage reference
Entity resolution
Decides that two records are the same real-world thing, with a survivorship rule and a confidence, so downstream joins are a decision rather than an assumption.
Match candidateSurvivorship ruleCluster
Data quality and contracts
Schema, freshness and volume expectations asserted at the boundary, so a bad load fails loudly here rather than quietly two layers later.
ContractExpectationViolation
Genealogy assemblythis domain adds
Reconstructs the lot-and-serial tree across execution and warehouse systems, so a defect can be traced to a supplier lot and forward to shipped units.
LotSerialGenealogy edge
What the engineer owns here
Material and location master alignment across systems that each think they own it.
Where this layer fails
Snapshot loads instead of change capture. It looks identical in a demo and it silently loses every intermediate state, which is exactly what an audit later asks you for.
03/07
Governance & semantics
Material master, BOM structures, governed metrics and the supply network graph.
Takes in
Replayable, resolved records with lineage from ingestion.
Hands on
Typed, classified, defined data an agent can be grounded on without inferring meaning.
SystemWhat it doesKey business entities
Business glossary
One agreed definition per term, owned by a named person, so the number an agent quotes means what the business means by it.
TermDefinitionOwnerSteward
Ontology and knowledge graph
The typed entities and relationships the domain actually has, so an agent can traverse "which customers are exposed to this" rather than guess from adjacent text.
EntityRelationshipNodeEdge
Semantic and metrics layer
Metrics defined once, in one place, with their filters and grain. Removes the class of error where the agent computed something plausible and wrong.
MetricDimensionGrainFilter
Classification and policy
Sensitivity labels and the purpose each classification permits, applied at the field level and inherited by everything downstream.
ClassificationPurposePolicy
Lineage and catalogue
Where a value came from and what it touched on the way. The layer that makes an answer defensible rather than merely correct.
Lineage edgeDatasetColumn
What the engineer owns here
One definition of OEE, on-time and inventory that everyone can be held to.
Where this layer fails
Letting the model infer meaning from column names. It will, it will be plausible, and nobody will notice until the number reaches a regulator or a board pack.
04/07
Retrieval & grounding
Retrieval over contracts, specifications, work instructions, quality records and service history.
Takes in
Typed, classified, defined data from the governance layer.
Hands on
Ranked, permission-trimmed, citable evidence scoped to the caller and the moment.
SystemWhat it doesKey business entities
Hybrid index
Vector and lexical retrieval together, because exact identifiers, codes and clause numbers are the thing semantic search is worst at.
ChunkEmbeddingTermScore
Chunking and metadata
Splits on the document’s own structure and carries effective dates, version and source into every chunk, so a retrieved passage knows when it was true.
ChunkEffective dateVersionSource reference
Security trimming
Applies the caller’s permissions inside the query rather than filtering results afterwards, so the model never sees what the user may not.
PrincipalEntitlementFilter
Graph retrieval
Traverses the knowledge graph for questions that are joins rather than similarity — exposure, genealogy, ownership, causation.
PathHopTraversal
Reranking and citation
Reorders candidates on relevance and returns the passage identifier behind every sentence, so the answer can be checked rather than trusted.
CandidateRankCitation
What the engineer owns here
Turning specifications and service notes into retrievable operational knowledge.
Where this layer fails
Filtering results after retrieval instead of constraining the query. The model has already seen the rows you removed, and it will use them.
05/07
Agent & orchestration
Planning exception agents, procurement agents, production monitoring and logistics exception agents.
Takes in
Ranked, permission-trimmed, citable evidence from retrieval.
Hands on
Actions taken or proposed, each with the evidence, the identity and the trace behind it.
SystemWhat it doesKey business entities
Agent runtime
Plans, calls tools and holds the loop. Where the step budget, the timeout and the stopping condition are enforced rather than hoped for.
TaskPlanStepTrace
Tool and MCP servers
Typed, permissioned tools with declared schemas. An agent’s real capability surface is this list, which is why the list is a security artefact.
ToolSchemaScopeInvocation
State and memory
Durable task state with checkpoints, so a run that dies mid-way resumes instead of restarting and re-doing side effects.
CheckpointSessionMemory record
Human-in-the-loop gates
Approval steps on the actions that need one, carrying enough context for the approver to actually decide rather than rubber-stamp.
ApprovalThresholdDecision
Workflow integration
Writes back into the systems people already work in, under a service identity with its own audit trail.
Work itemWrite-backActor
What the engineer owns here
Reaching the execution layer — triggering real workflow, not producing another report.
Where this layer fails
An agent that acts under a service account rather than on behalf of the user. It will eventually do something the requesting user had no right to do, and the log will not show it.
06/07
Guardrails & operational control
Commitment limits, safety boundaries on shop-floor advice and change control.
Takes in
Proposed actions and drafted responses from the agent layer.
Hands on
Permitted, grounded, logged output — or a refusal with a stated reason.
SystemWhat it doesKey business entities
Policy enforcement
The rules that decide whether an action is permitted at all, evaluated before the action and independently of the model that proposed it.
PolicyDecisionObligation
Input and output screening
Injection detection on the way in, and on the way out the checks for leakage, unsupported claims and content the domain forbids.
PromptResponseViolation
Groundedness checking
Verifies each assertion resolves to retrieved evidence, and fails the response rather than shipping the sentence that does not.
AssertionSupportVerdict
Model risk management
Model inventory, intended use, validation evidence and the sign-off that lets a model be used for a purpose. Not optional in a regulated estate.
ModelIntended useValidationApproval
Audit log
Immutable record of what was asked, retrieved, decided and done — the artefact a reviewer reads when they do not take your word for it.
Audit entryActorEvidence reference
What the engineer owns here
Hard limits on anything that commits spend, capacity or a customer promise.
Where this layer fails
Guardrails that only run on the prompt. Most real failures are on the output side: the unsupported sentence, the leaked field, the claim nobody approved.
07/07
Evaluation & observability
Plan-quality evals, OEE impact tracking, exception precision and cost per decision.
Takes in
Permitted, grounded, logged output from the guardrail layer.
Hands on
Measured quality, cost and latency — and the evidence to change any of the three.
SystemWhat it doesKey business entities
Evaluation harness
Held-out sets and graded runs on every change, so a prompt edit is a measured change rather than a hopeful one.
Eval setCaseScoreBaseline
Regression gate
Blocks a release when quality drops, in CI, on the same evidence for everyone. The difference between a system and a demo.
ThresholdRunGate decision
Tracing
End-to-end spans across retrieval, model and tool calls, so a bad answer can be opened and read rather than argued about.
TraceSpanLatencyToken count
Cost and capacity
Cost per task and per tenant, against throughput and quota. The number that decides whether the pilot can become the rollout.
Cost per taskQuotaThroughput
Drift and feedback
Watches quality against production traffic rather than the test set, and routes real corrections back into the eval suite.
Drift signalFeedbackIncident
What the engineer owns here
Attributing an OEE or service improvement to the agent rather than to the weather.
Where this layer fails
Evaluating on steady-state production. The cases that matter are changeovers, ramp-ups and the disrupted week, which is when the model has least support.
Typical tech stack
Models
- Frontier models for reasoning and drafting
- Time-series and optimisation models
- Vision models for inspection
Grounding
- Contract and specification retrieval
- Work instruction index
- Service history mining
Agents
- Planning and procurement agents
- MCP over ERP and MES
- Simulation and solver tools
Data
- Event streaming from shop floor
- Material master harmonisation
- Supply network graph
Assurance
- Plan feasibility validation
- Impact attribution
- Exception precision tracking
Runtime
- Cloud plus edge at the plant
- Approval workflow
- Commitment guardrails
Constraints that shape the design
Feasibility is non-negotiable
A recommendation that violates a real constraint destroys trust permanently. Validate against the constraint model before proposing.
Master data across systems
ERP, MES and WMS each believe they own material identity. Reconciling them is usually the critical path.
Reach the execution layer
If the output is a report, it will be ignored. The agent has to trigger the workflow that already exists.
Plant autonomy
Plants operate differently for real reasons. A model that works at one site needs local calibration, not enforcement.
04
Part 3 · Getting it done
Where the business and the technology meet
Twelve use cases across plan, source, make, quality, deliver and service — each tied to a chain stage and the architecture layer that does the work.
Use cases, mapped to the chain and the architecture
Filter by stage or by earned autonomy. Selecting a use case jumps the value chain and the architecture to the stage and layer it depends on.
Value chain stage
Autonomy earned
12 use cases
No use cases match that combination.
The delivery playbook
How an industrial deployment actually runs.
01
Week 1 — pick one plant, one line
Site-level variation is real. Prove it where you can stand next to the machine, then generalise.
02
Week 2 — reconcile the master data
Get ERP, MES and WMS to agree on material and location identity. This is usually the critical path.
03
Weeks 3-5 — build the constraint model
Encode the real constraints with the planner and the plant. An infeasible recommendation costs you the deployment.
04
Weeks 6-8 — reach the execution layer
Wire the agent into the workflow that already exists in ERP. A recommendation that needs re-keying will not be used.
05
Weeks 9-12 — attribute and scale
Measure impact against a comparable line or period, then hand over with a site-calibration guide.
The curriculum
8 modules, 80 taught hours, 40 hands-on labs and 8 assessments — every lab provisioned and graded by the SCIKIQ Agentic AI Playground. Open a module to see its labs.
01The SCDAI delivery model in industrial operationsLanding at a plant where credibility is earned on the floor, not in a workshop.10 hrs · 5 labs
Delivery mandateSite selectionValue sizing on OEE and serviceStakeholder mapPlant autonomyThin-slice scoping
01
Site selectionScore plants on data maturity, willingness and value at stake.
02
Value sizingQuantify an OEE point and a service-level point in cash terms.
03
Gemba discoveryStructure a shop-floor discovery that produces a decision inventory.
04
Stakeholder mapMap plant manager, planner, quality, IT and OT.
05
Thin sliceScope a one-line, one-decision slice with an execution-layer landing point.
Assessment 01 — Operating model & scoping
02Industrial data foundationsERP, MES, WMS and EAM — and the master data that has to reconcile them.10 hrs · 5 labs
ERP data structuresBOMs and routingsMES event dataMaterial masterWarehouse and transport dataData quality
01
ERP extractionExtract orders, inventory and BOM structures with lineage.
02
MES eventsIngest downtime and quality events at line speed.
03
Material harmonisationReconcile material identity across ERP, MES and WMS.
04
Supply network graphBuild a graph of suppliers, sites, materials and routes.
05
Quality gatesDetect master data breakage before it reaches a recommendation.
Assessment 02 — Industrial data landscape
03Context engineering for operationsMaking recommendations that a planner or plant manager can verify in thirty seconds.10 hrs · 5 labs
Explanation designConstraint communicationStructured recommendationsUncertaintyRefusalContext budget
01
Trade-off explanationExplain a planning trade-off in terms the planner already uses.
02
Constraint groundingEnsure every recommendation states the binding constraint.
03
Recommendation schemaEmit structured options with impact and feasibility flags.
04
RefusalMake the agent decline when the constraint model is stale.
05
Context budgetReduce context cost on a high-frequency monitoring agent.
Assessment 03 — Context engineering
04Grounding on operational knowledgeContracts, specifications, work instructions and service history as retrievable context.10 hrs · 5 labs
Contract retrievalSpecification retrievalWork instructionsService note miningCitationsRetrieval evals
01
Contract extractionExtract pricing and rebate terms with clause citations.
02
Work instruction retrievalAnswer shop-floor questions from the correct instruction revision.
03
Service miningExtract failure modes and fixes from unstructured service notes.
04
Citation contractCite the document and revision behind every operational answer.
05
Retrieval evalsMeasure recall on labelled operational questions.
Assessment 04 — Retrieval & grounding
05Agent design for planning and executionException agents, solver tools and reaching the transaction rather than the report.10 hrs · 5 labs
Workflow vs agentSolver and simulation toolsException designExecution-layer integrationMulti-agent flowsContainment
01
Exception agentSurface planning exceptions with feasible, ranked options.
02
Solver toolGive the agent an optimisation tool and validate feasibility.
03
Execution landingTrigger a real ERP workflow from an approved recommendation.
04
Agent fleetCoordinate planning, procurement and logistics agents.
05
ContainmentCap commitment authority for spend and capacity.
Assessment 05 — Agent design & execution
06Integrating with ERP and the shop floorSAP and MES integration, edge deployment and the reality of a plant network.10 hrs · 5 labs
MCP over ERPMES integrationEdge deploymentLatency budgetsIdentityError handling
01
MCP over ERPExpose planning and order data as typed tools.
02
MES integrationConsume shop-floor events and write structured downtime reasons.
03
Edge inferenceRun inspection inference at the line with bounded latency.
04
Latency budgetMeet a line-speed latency requirement end to end.
05
Error handlingEnsure a retry never duplicates a production order.
Assessment 06 — Systems integration
07Safety, security and operational riskWhere industrial AI can hurt someone, and how to make sure yours cannot.10 hrs · 5 labs
Shop-floor safety boundariesOT securityChange controlOWASP LLM Top 10Injection via documentsFailure analysis
01
Safety boundaryEnforce that the agent cannot influence a safety function.
02
OT threat modelThreat-model plant-floor integration and prove no control path.
03
Injection defenceDefend against instructions embedded in a supplier document.
04
Change controlDefine what agent change requires plant re-validation.
05
Failure analysisAnalyse the consequences of a confidently wrong recommendation.
Assessment 07 — Safety, security & risk
08Production, attribution and costProving the improvement was yours, and keeping it across sites.10 hrs · 5 labs
Eval suitesImpact attributionSite calibrationDriftCost per decisionHandover
01
Eval gateGate deployment on plan feasibility and recommendation quality.
02
AttributionDesign a comparison that credibly attributes an OEE gain.
03
Site calibrationAdapt a model to a second site and document what changed.
04
Cost per decisionMeasure and reduce cost per recommendation at scale.
05
HandoverProduce runbook, calibration guide and owner assignment.
Assessment 08 — Production & operations
Two capstones — 20 hours
End-to-end agent design, development, deployment and testing, 10 hours each. Reviewed by a Senior SCDAI Engineer against a published rubric.
Capstone A — Planning exception agent
Build an agent that surfaces planning exceptions with feasible, constraint-validated options and lands an approved decision in the real ERP workflow.
Capstone B — OEE loss diagnosis agent
Build an agent that correlates downtime, changeover, quality and maintenance data to explain OEE loss with evidence, and measure its diagnostic accuracy against engineer labels.
What you deliver, and how it is marked
- A running agent, deployed on the program's target stack, reachable by a reviewer.
- An architecture note: boundaries, tools, autonomy level and the trade-offs you took.
- A versioned eval set with a baseline and your final scores.
- A trace walkthrough of one success and one deliberate failure.
- Cost per task, measured — not estimated.
- A runbook and rollback path a client team could operate on Monday.
Problem fit15%Is this a decision worth automating, scoped to a defensible boundary?
Architecture20%Right pattern for the constraints; autonomy earned, not assumed.
Grounding & accuracy20%Measured groundedness against the eval set, with the failures named.
Safety & compliance20%Domain controls actually enforced, and demonstrably so.
Operability15%Traces, runbook, rollback, cost — could someone else run this?
Communication10%Can a client executive follow the reasoning in ten minutes?
One lab, start to finish
Every lab in this program follows the shape below. This is lab 05 in full — the brief you are given, the environment that is provisioned for you, the code you start from and the assertions that decide whether you passed.
Module 05 · Agent design for planning and execution
Feasible-only recommendations
Never propose a plan that violates a real constraint.
The brief
Build a planning exception agent over a seeded plant. Constraints include changeover matrices, material availability, shift calendars and a sequencing rule the planner will tell you about only if you ask. Every recommendation must validate before it is shown.
Provisioned for you
- ERP extract: orders, BOMs, routings
- Constraint model with changeover matrix
- Solver exposed as a tool
- 30 planning scenarios, 8 infeasible by design
You start from this
def propose(scenario, solver, constraints):
"""Validate BEFORE returning. An infeasible recommendation shown once
costs more trust than ten it never made."""
options = solver.solve(scenario)
raise NotImplementedErrorWhat the grader asserts
- Zero infeasible options returned across all 30 scenarios
- The 8 impossible scenarios return "no feasible option" with the binding constraint named
- Each option states the constraint that limits it
- Recommendations land as an ERP workflow trigger, not a report
- Agent refuses when the constraint model is older than the shift calendar
05
Assessment
Assessed, not attended
Every module ends with a timed, randomised assessment delivered through the SCIKIQ Agentic AI Playground. The certificate requires a pass on all of them plus two reviewed capstones.
Assessment blueprint
What each test covers, how long it runs, and how many items are currently in the versioned bank behind it.
| # | Assessment & coverage | Items | Time | In bank |
|---|---|---|---|---|
| 01 | Operating model & scopingDelivery mandateSite selectionValue sizingPlant dynamics | 25 | 35 min | 4 |
| 02 | Industrial data landscapeERP structuresMES eventsMaterial masterNetwork modelling | 25 | 35 min | 4 |
| 03 | Context engineeringExplanation designConstraint communicationStructured outputsRefusal | 25 | 35 min | 4 |
| 04 | Retrieval & groundingContract retrievalWork instructionsService miningRetrieval metrics | 25 | 40 min | 4 |
| 05 | Agent design & executionException designSolver integrationExecution layerContainment | 25 | 40 min | 4 |
| 06 | Systems integrationMCP over ERPMES integrationEdge and latencyResilience | 25 | 35 min | 4 |
| 07 | Safety, security & riskSafety boundariesOT securityLLM securityChange control | 30 | 45 min | 4 |
| 08 | Production & operationsEval gatingImpact attributionMulti-site scalingEconomics and handover | 30 | 45 min | 4 |
Single best answerOne defensible option among plausible distractors drawn from real field mistakes.
Scenario judgementA client situation with constraints; you pick the action a Senior SCDAI Engineer would take.
Architecture selectionGiven non-functional requirements, choose the pattern and justify the trade-off.
Failure diagnosisA trace, an eval report or a cost curve — identify the root cause.
- Each module ends with a timed assessment; the program certificate requires a pass on every one.
- Items are drawn at random from a versioned bank, so no two learners sit an identical paper.
- Pass mark is 70%. Two retakes are included, each with a fresh draw.
- Scenario items carry double weight — they are the ones that predict field performance.
- Capstones are reviewed against a published rubric by a Senior SCDAI Engineer, not auto-graded.
Try four questions from the bank
Real items, drawn from 32 in this program's bank — weighted toward the scenario and diagnosis types, because those are the ones that predict field performance. Instant feedback, nothing saved.
Four sample items — one attempt each, then the reasoning is shown.
Q1Why start at one plant and one line?
Site variation is genuine. Proving it locally, then generalising with calibration, is what works.
Q2ERP, MES and WMS each maintain material identity. What is the consequence?
Master data reconciliation is usually the critical path in industrial deployments, and it is chronically underestimated.
Q3A planner rejects a recommendation as infeasible. What has failed?
Infeasibility is the fastest way to lose a planning deployment. Validate against constraints before proposing.
Q4Work instructions have revisions. What must retrieval guarantee?
Newest is not always correct. Applicability, like wording edition in insurance, is the real requirement.
06
Career & outcomes
What this makes you, and where it goes next
The same ladder whichever specialisation you enter through — what changes is the domain you go deep in. Below: how the program is delivered, the skills it moves, the roles it leads to, and the specialisations closest to this one.
Available two ways
The same labs, assessments and capstones, delivered to an enterprise cohort or to individual professionals.
B2B
Direct to enterprise
Cohorts of 20 to 2,000+ on your own tenancy, with your data patterns and your cloud. Skill-gap baselining up front, per-team mastery reporting throughout, and capstones scoped against your real backlog so the output is deployable work.
- Private cohort on your tenancy
- Baseline skills assessment before kick-off
- Capstones scoped to your backlog
- Per-team mastery dashboards for L&D
- Optional Senior SCDAI Engineer architecture reviews
B2C
Direct to individual professionals
The same labs, assessments and capstones for individual engineers and analysts, run on shared infrastructure with a fixed cohort calendar. You leave with a graded portfolio, not a certificate of attendance.
- Fixed-date cohorts, self-paced within them
- Same graded labs and item bank
- Portfolio-ready capstones you own
- Certificate on a pass in every module
- Community and office-hours access
Skill matrix
Find your row and aim one column right. The Playground scores you against this after every module.
| Skill | Beginner | Intermediate | Advanced |
|---|---|---|---|
| Operations fluency | Knows the terms. | Maps plan-source-make-deliver to decisions. | Sizes OEE and service value and designs the change. |
| Constraint modelling | Reads a plan. | Validates recommendations against constraints. | Designs constraint models plants trust. |
| Execution integration | Produces a report. | Triggers workflow in ERP. | Designs agents that operate inside transactions. |
| Context engineering | Writes clear prompts. | Structures retrieval, tools and state deliberately. | Designs context strategy for reliability and cost at scale. |
| Retrieval & grounding | Builds basic vector search. | Tunes chunking, hybrid search and reranking. | Designs graph + vector grounding with measured recall. |
| Agent orchestration | Runs a single tool-calling agent. | Builds supervised multi-step and multi-agent flows. | Designs autonomy boundaries and failure containment. |
| Tool & system integration | Calls a documented API. | Writes an MCP server over a system of record. | Designs a least-privilege tool estate across systems. |
| Evaluation | Eyeballs outputs. | Builds labelled eval sets and regression gates. | Runs online evals with drift and judge calibration. |
| Observability & cost | Reads logs. | Traces runs, tracks tokens and latency. | Owns cost per task and capacity planning in production. |
| Security & guardrails | Adds output filters. | Mitigates the OWASP LLM Top 10 in a build. | Threat-models an agent estate and proves controls. |
| Client delivery | Takes notes in a workshop. | Runs discovery and scopes a thin slice. | Owns the account technically, from scope to handover. |
Where it takes you
The SCDAI ladder is the same whichever specialisation you enter through — what changes is the domain you go deep in.
01
Data and AI EngineerEntry
Python, LLM APIs, first RAG build, prompt and context basics.
02
Associate Data and AI EngineerPractitioner
Client-facing discovery, tool integration, agent build under supervision.
03
SCIKIQ Certified Data and AI EngineerSpecialist
Owns a deployment end to end — scoping, build, evals, handover.
04
Senior Data and AI Engineer / Domain LeadSenior
Multi-agent architecture, regulated deployments, reference patterns for the practice.
05
AI Architect / Practice LeadLeadership
Portfolio of accounts, delivery standards, hiring and capability strategy.
Terms worth knowing
- OEE
- Overall Equipment Effectiveness — availability × performance × quality.
- S&OP
- Sales and Operations Planning — the cross-functional balancing cadence.
- MES
- Manufacturing Execution System — the shop-floor system of record.
- BOM
- Bill of Materials — the component structure of a product.
- OTIF
- On-Time In-Full — the headline delivery performance measure.
- Should-cost
- A modelled estimate of what a purchased item ought to cost.
- First-pass yield
- The proportion of units produced correctly without rework.
- Changeover
- The time and work to switch a line from one product to another.
- SCDAI
- SCIKIQ Certified Data and AI Engineer — the credential this program awards.
- SCDAI engineer
- An engineer embedded with the customer who turns their constraints into a working system.
- Thin slice
- The narrowest end-to-end path through a workflow that still produces business value.
- Agentic RAG
- Retrieval where the agent decides when, what and how often to retrieve.
- MCP
- Model Context Protocol — open standard connecting an agent to tools and data.
- A2A
- Agent2Agent — protocol for agent-to-agent delegation across vendors.
- Eval set
- A labelled, versioned set of cases an agent must pass before release.
- LLM-as-judge
- Using a model to score outputs, calibrated periodically against human ratings.
- Guardrail
- An enforced constraint on an input, output or action — not a prompt request.
- Human-in-the-loop
- A required human approval step before a consequential action executes.
- Cost per task
- Total token, tool and compute cost of one completed unit of work — the engineer's unit economic.
- Golden path
- The reference architecture a practice standardises on so deployments stay reviewable.
- Handover
- The point at which the client can operate, evaluate and change the system without the engineer.
Related programs
Skill a team, or join a cohort
B2B cohorts run on your tenancy with capstones scoped to your backlog. B2C cohorts run on a fixed calendar.
07
Enrolling
What you need, what you get, and what it costs
Stated plainly enough to rule yourself in or out without a sales call: the prerequisites, how the program runs, exactly what the credential is worth, and the questions everyone asks.
Before you start
Stated plainly so you can rule yourself in or out without a sales call. Nothing here is a formal qualification — it is what the first lab assumes you can already do.
Technical
You should already be able to do these
- Comfortable in Python: functions, typing, virtual environments, reading a stack trace
- Have called an LLM API and handled a structured response
- Git, the command line, and reading someone else’s code without panic
Domain
What we assume, and what we teach
- Familiarity with the idea of a plan, an order and a bill of materials
- Any shop-floor or planning exposure shortens the on-ramp considerably
Time and access
What the program asks of your week
- A machine that can run a browser and a terminal — labs run in the Playground, not locally
- Roughly 8 hours a week for the taught weeks, plus 20 hours for the two capstones
How it runs
Cohort dates and pricing are confirmed on enquiry rather than printed here, because both move with the intake.
- Format
- Self-paced within a fixed cohort window, delivered in the SCIKIQ Agentic AI Playground
- Taught hours
- 80 hours across 8 modules
- Capstones
- 20 hours — two end-to-end builds, reviewed not auto-graded
- Typical duration
- 10–12 weeks at around 8 hours a week
- Level
- Intermediate → Advanced
- Built for
- Engineers and analysts deploying AI across manufacturing and supply chain
- Next cohorts
- — confirmed on enquiry
- Pricing
- — confirmed on enquiry
The credential
The credential is awarded per specialisation, so it names the domain or stack you were assessed in rather than claiming general competence. On this program the badge reads SCDAI — Manufacturing & Supply Chain.
How it is earned
- A pass at 70% or above on every module assessment — not an average across them
- Both capstones reviewed against the published rubric by a Senior SCDAI Engineer
- All labs submitted and machine-graded green
- Awarded as
- A verifiable digital badge with a public credential record
- Valid for
- Two years from award
- Recertification
- One current-year assessment plus a refreshed capstone
- Retakes
- Two included per assessment, each with a fresh item draw
- Named specialisation
- The badge states the domain or stack, e.g. SCDAI — Healthcare
- Verification
- Third parties confirm a holder from the credential record, no login required
A certificate that cannot be checked is decoration. Every award resolves to a record showing the specialisation, the award date and the assessments passed.
The certificate you leave with
This is the credential the program awards, shown exactly as it is issued — with the specialisation named, the assessment record attached and a verification link anyone can check without an account.
SCIKIQ Data Academy
Certificate of Professional Certification
This is to certify that
Your name
has been assessed and certified as
SCIKIQ Certified Data and AI Engineer
Manufacturing & Supply Chain
- Specialisation
- Manufacturing & Supply Chain
- Assessments passed
- 8 of 8, each at 70% or above
- Labs graded
- 40 of 40 machine-graded green
- Capstones reviewed
- 2, against the published rubric
- Taught hours
- 80 across 8 modules, plus 20 capstone hours
AwardedOn completion
Valid untilTwo years from award
Credential IDSCDAI-MANUFACTURIN-<issued>
Senior SCDAI Engineer, Certification Board
- 01The specialisation is on the certificate
Not “Data and AI Engineer” but the domain or stack you were actually assessed in. A general claim would be a weaker one.
- 02The assessment record travels with it
Modules passed, labs graded and both capstones reviewed — so the credential states what was measured rather than that you attended.
- 03Anyone can verify it, without a login
The credential ID resolves to a public record showing the specialisation, the award date and the assessments passed.
Share it on LinkedIn
Add it to your LinkedIn profile in one step. The link pre-fills the certification fields from the credential record, so the entry on your profile matches the record a reader can check.
Add to LinkedIn profile The button is live on your real certificate; here it opens LinkedIn pre-filled with this specialisation so you can see exactly what the profile entry will say.A certificate that cannot be checked is decoration. Every award resolves to a record showing the specialisation, the award date and the assessments passed.
Questions we get asked
The objections that come up in every conversation about this program, answered without the brochure voice.
Is this a course or a certification?
Both, and the second is the point. Eight timed assessments and two reviewed capstones stand between you and the credential, so a pass means someone measured the skill rather than recorded your attendance.
How is it self-paced without falling apart?
Every lab is provisioned, graded and unblocked by the Playground rather than by an instructor. That is what lets a cohort of 2,000 cost the same faculty time as a cohort of 20 — and why you are never waiting on someone to mark your work.
What if I fail an assessment?
Two retakes are included per assessment, each drawing a fresh item set from the bank, so a retake is a genuinely new paper rather than the same questions again.
Do I need my own cloud account?
For the domain specialisations, no — labs run in provisioned sandboxes. For the four tech-stack programs you will want access to that platform, since deploying into a real subscription is much of the point.
Can we run this privately for our team?
Yes. B2B cohorts run on your own tenancy with your data patterns, a skills baseline before kick-off, per-team mastery reporting, and capstones scoped against your actual backlog so the output is deployable work rather than an exercise.
Which specialisation should I take?
Take the domain you deploy into. If you move across industries, take a tech-stack program instead and pick up domain context on the engagement. The chooser on the programs page will narrow it.
How current is the material?
The trends, platform capabilities and regulatory positions are reviewed each quarter, and every external claim on these pages links to its source so you can check the date yourself.
What do I actually leave with?
A graded portfolio: forty machine-graded labs, two reviewed end-to-end agent builds with measured evaluation and cost per task, and a verifiable credential naming your specialisation.
Still deciding?
Tell us the systems you deploy into and we will say plainly whether this specialisation is the right one — or which of the 21 is.