AI for Process Optimisation Training Course

Participants can document an as-is process at a level suitable for improvement analysis.

• BPMN 2.0 events, tasks, gateways, pools and lanes
• As-is workflow mapped from trigger through completion and exception handling
• Decision points distinguished from tasks, approvals and automated actions
• Handoffs documented with accountable roles, systems and information requirements
• Happy-path and exception-path flows separated within one process model
• Process boundaries tested against upstream dependencies and downstream customer outcomes

Participants can assess whether process data is fit for analysis and AI development.

• Event-log structure: case identifier, activity, timestamp, resource and attributes
• Data lineage traced from source systems through reports and analytical datasets
• Missing timestamps, duplicate events and inconsistent activity labels identified
• Master-data quality assessed for customers, products, suppliers and service categories
• Data dictionary specifying field definitions, business rules and permitted values
• Privacy classification applied to personal, confidential and restricted operational data

Participants can build a process baseline using service, cost, flow and quality measures.

• Cycle time, touch time, wait time and lead time distinguished
• First-pass yield, rework rate and defect opportunities calculated
• Backlog ageing profiles exposing stalled cases and service-risk concentrations
• Operational KPI tree linking process measures to strategic business outcomes
• Baseline dashboard combining volume, timeliness, quality, cost and risk indicators
• Control limits separating common-cause variation from special-cause variation

Participants can identify flow losses and constraints before applying AI solutions.

• Eight wastes identified within transaction, service and knowledge-work processes
• Value-stream map quantifying queues, rework, handoffs and non-value-added effort
• Theory of Constraints used to locate system bottlenecks and protective capacity
• Little's Law applied to backlog, throughput and average cycle time
• Root-cause analysis using five whys, fishbone diagrams and cause verification
• Future-state principles separating simplification opportunities from AI opportunities

Participants can interpret variation patterns and identify stable improvement opportunities.

• Run charts revealing trends, shifts, cycles and unusual process behaviour
• Control charts selected for continuous, count and proportion-based measures
• Process capability indices Cp and Cpk interpreted against specification limits
• Pareto analysis ranking defect categories, delay reasons and exception types
• Stratification by channel, product, customer segment, location and team
• Correlation findings distinguished from causal explanations and intervention evidence

Participants can interpret process mining outputs to identify improvement priorities.

• Process discovery reconstructing actual workflows from timestamped event logs
• Conformance checking comparing observed paths against approved BPMN process models
• Variant analysis identifying high-cost, delayed and non-compliant process paths
• Throughput analysis locating wait states, rework loops and repeated approvals
• Organisational mining examining handoffs, workload distribution and collaboration patterns
• Process mining findings translated into operational hypotheses and improvement actions

Participants can formulate AI use cases that address a defined process decision.

• Decision inventory listing recurring judgments, predictions and allocation choices
• Use-case categories: prediction, classification, recommendation, optimisation and generation
• AI suitability screened against data availability, repeatability and decision frequency
• Rules-based automation separated from machine-learning and generative-AI requirements
• Opportunity statements linking model output to named operational actions
• Failure modes considered before selecting an AI-enabled intervention

Participants can rank AI opportunities using transparent selection criteria.

• Use-case scorecard combining value, feasibility, risk, urgency and strategic fit
• Benefit hypotheses quantified through labour, capacity, quality and service drivers
• Cost-to-serve analysis estimating economic impact of process delays and errors
• Data maturity assessed across accessibility, quality, coverage and historical depth
• Risk register documenting customer harm, compliance exposure and operational failure
• Prioritisation matrix separating quick wins, foundations, pilots and strategic bets

Participants can specify predictive models that support process interventions.

• Prediction targets defined for delays, demand, defects, churn and escalations
• Feature design using transaction history, process state and contextual variables
• Classification models explained through precision, recall, F1 score and thresholds
• Regression models assessed with MAE, RMSE and forecast bias
• Time-series forecasting applied to workload, arrivals and staffing requirements
• Prediction outputs connected to triage, scheduling and proactive customer contact

Participants can define optimisation models for operational allocation and scheduling decisions.

• Objective functions balancing service, cost, quality, risk and utilisation
• Decision variables specified for staffing, routing, sequencing and inventory allocation
• Constraints modelled for skills, capacity, policy, budget and service commitments
• Linear programming principles applied to resource allocation decisions
• Scenario analysis comparing operating choices under changing demand assumptions
• Recommendation logic tested against practical process constraints and exceptions

Participants can design bounded generative AI applications for knowledge-intensive process tasks.

• Generative AI use cases for summarisation, drafting, extraction and classification
• Prompt templates structured with task, context, source, format and constraints
• Retrieval-augmented generation grounding outputs in approved internal knowledge sources
• Output validation rules preventing unsupported statements and incorrect classifications
• Human review points assigned according to impact, ambiguity and regulatory sensitivity
• Prompt-injection and confidential-data risks addressed through workflow controls

Participants can specify a workflow that combines AI, rules and human judgment.

• Decision automation patterns: straight-through processing, triage, assist and escalation
• Business rules documented separately from probabilistic model recommendations
• Confidence thresholds determining automatic action, review queues and exception handling
• RPA handoffs designed between legacy systems, AI services and human teams
• Service-level agreements defined for automated decisions and review turnaround
• Fallback procedures maintaining operations during model or integration failure

Participants can evaluate whether an AI intervention improves process performance safely.

• Training, validation and test datasets separated to prevent performance overstatement
• Confusion matrices interpreted for operational classification decisions and error costs
• Threshold selection aligned with false-positive and false-negative consequences
• A/B test design comparing AI-supported workflow performance against baseline operations
• Pilot success criteria defined for adoption, quality, throughput and customer outcomes
• Error analysis examining cases where model recommendations fail operationally

Participants can establish controls for accountable and auditable AI-supported decisions.

• Model risk classification based on decision impact, autonomy and customer exposure
• Bias testing across protected groups, service segments and geographic populations
• Explainability requirements matched to users, regulators, operators and affected customers
• Human-in-the-loop responsibilities defined through approval and escalation matrices
• Model documentation recording purpose, data, limitations, owners and review dates
• Audit trail requirements covering inputs, recommendations, overrides and final decisions

Participants can plan controlled deployment and ongoing performance monitoring for operational AI.

• Deployment options compared for batch scoring, real-time APIs and embedded workflows
• Model monitoring dashboard tracking accuracy, drift, latency and business outcomes
• Data drift and concept drift distinguished using operational examples
• Retraining triggers established from performance deterioration and process changes
• Version control applied to datasets, model artefacts, prompts and business rules
• Incident management process defining containment, investigation, approval and recovery

Participants can prepare teams and operating roles for AI-enabled process changes.

• Stakeholder impact assessment identifying changed tasks, authority and accountability
• Future-role design separating automated work, assisted work and expert judgment
• RACI matrix assigning ownership for data, models, workflows and exceptions
• Standard operating procedures revised for AI recommendations and manual overrides
• Training plan addressing new decisions, escalation paths and quality controls
• Adoption measures tracking utilisation, override rates, trust and compliance

Participants can develop an investment case and staged delivery plan for a priority use case.

• Benefits case connecting intervention mechanisms to measurable financial and service results
• Total cost estimate covering data, integration, licences, training and support
• Sensitivity analysis testing benefit assumptions, adoption rates and delivery delays
• Roadmap sequenced through discovery, pilot, deployment, monitoring and scale-up
• Dependency map identifying process redesign, data remediation and technology prerequisites
• Executive decision paper presenting recommendation, risks, controls and required approvals

Participants can integrate process, data, AI and governance elements into a review-ready design.

• As-is and target-state process models aligned with proposed decision automation
• Data readiness findings translated into remediation actions and ownership
• Use-case scorecard reconciled with business case, risk register and roadmap
• Model specification linked to operational actions, thresholds and human reviews
• Control plan connecting process KPIs, model measures and escalation triggers
• Review criteria testing feasibility, value, accountability and implementation readiness

Participants can build, review and present a complete AI-enabled process optimisation blueprint for a selected operational process.

• Process baseline presenting flow, performance gaps, bottlenecks and customer impacts
• Evidence-backed use case defining decision, AI method, users and actions
• Target workflow integrating automation, model recommendations and human oversight
• Data plan specifying sources, quality gaps, access controls and remediation
• Implementation roadmap detailing pilot scope, metrics, governance and change actions
• Executive presentation defending value assumptions, risks, controls and investment request