AI Strategy and Governance Training Course

Participants can formulate AI ambitions and prioritise use cases against value, feasibility, risk and readiness.

• Strategic intent translated into AI ambition, scope, and value themes
• Use-case portfolio scored for value, feasibility, risk, and data readiness
• Value hypothesis defined with baseline, benefit mechanism, owner, and metric
• Process maps expose decision bottlenecks, handoffs, exceptions, and automation candidates
• Prioritisation matrix compares quick wins, strategic bets, and mandatory controls
• Benefits realisation plan assigns measurement cadence, owners, assumptions, and dependencies

Participants can assess the data, knowledge, ownership and access conditions required for dependable AI services.

• Data product ownership defined for model training, retrieval, and monitoring
• Data quality dimensions assessed: accuracy, completeness, timeliness, consistency, and validity
• Lineage records connect source data, transformations, features, and model outputs
• Personal data minimisation applied to training, prompting, and retention decisions
• Knowledge retrieval design compares vector search, metadata filters, and access controls
• Data access controls align role permissions with approved AI purposes

Participants can evaluate technical delivery options and define architecture decisions for AI initiatives.

• Reference architecture connects data sources, models, orchestration, interfaces, and monitoring
• Build, buy, and partner choices tested against capability, cost, and control
• Cloud and on-premises deployment patterns assessed for sensitive workloads
• Retrieval-augmented generation designed to ground answers in approved knowledge
• API integration boundaries defined between AI services and core systems
• Model selection criteria compare performance, latency, cost, explainability, and licensing

Participants can specify practical controls for generative AI applications, AI agents and human review.

• Prompt patterns specify role, task, context, constraints, and output format
• Grounding controls reduce unsupported answers through retrieval, citations, and abstention
• Agent workflows separate planning, tool use, approvals, and audit logs
• Evaluation sets test factuality, relevance, safety, and task completion
• Red-team scenarios probe prompt injection, data extraction, and harmful outputs
• Human-in-the-loop design assigns review thresholds, escalation, and override authority

Participants can establish governance structures, policies, records and decision rights for AI portfolios.

• ISO/IEC 42001 management-system requirements mapped to accountable AI governance
• AI policy defines permitted uses, prohibited uses, and approval routes
• Three-lines model assigns oversight, challenge, control testing, and assurance
• Model inventory records purpose, owner, version, data, and risk rating
• RACI matrix clarifies executive sponsorship, product accountability, and technical responsibilities
• Change-control gates govern releases, retraining, incidents, and retirement decisions

Participants can identify AI risks, select controls and document residual risk for accountable approval.

• NIST AI RMF functions organise govern, map, measure, and manage activities
• Impact assessment identifies affected groups, harms, likelihood, and severity
• Bias testing distinguishes sampling, measurement, proxy, and allocation bias
• Explainability requirements matched to user needs, decisions, and regulatory scrutiny
• Risk register links controls, residual risk, owners, and review dates
• Incident response covers harmful output, drift, breach, and vendor failure

Participants can translate legal and regulatory obligations into AI documentation, oversight and assurance activities.

• EU AI Act risk categories assessed against intended system purpose
• High-risk system obligations translated into documentation, oversight, and quality controls
• Privacy impact assessments address lawful basis, transparency, retention, and rights
• Copyright and intellectual-property risks assessed for training data and generated content
• Records support audit trails, contestability, disclosures, and regulator enquiries
• Cross-border deployment considers data residency, transfers, suppliers, and local obligations

Participants can organise AI delivery teams, assess suppliers and plan workforce adoption measures.

• Target operating model defines central, federated, and hybrid AI responsibilities
• Product teams align domain experts, engineers, risk partners, and users
• Vendor due diligence tests model provenance, security, resilience, and subcontractors
• Contract clauses address data use, audit rights, indemnities, and exit
• AI literacy programme differentiates user, manager, developer, and oversight training
• Adoption metrics track usage, workarounds, satisfaction, and realised benefits

Participants can create an AI roadmap that links investment decisions, delivery gates, governance milestones and executive reporting.

• Roadmap sequences foundations, pilots, scale decisions, and governance milestones
• Investment case combines implementation cost, operating cost, benefits, and risk
• Pilot charters define hypothesis, scope, success measures, and exit criteria
• Stage-gate reviews decide iterate, scale, pause, or retire initiatives
• Executive dashboard reports portfolio value, risk exposure, delivery status, and decisions
• Personal application plan specifies stakeholder conversations, first actions, and evidence