Glossary

Accountability is the obligation to another party for fulfilling a responsibility. It includes informing and explaining how that responsibility was discharged, with enforceable consequences where it was not. Accountability is not the same as performing the operational work.

AI governance connects purpose, roles, decision rights, policies, controls and evidence so that AI systems are used throughout their lifecycle in line with strategy, risk tolerance, values and applicable requirements. It links technical choices to organisational accountability and defines how systems are selected, developed, acquired, evaluated, operated, monitored, changed and retired. It is not a single committee or one-off approval.

AI risk management is a continuous, contextual process for identifying, analysing, evaluating, treating and monitoring the risks of an AI system throughout its lifecycle. NIST structures this voluntarily through Govern, Map, Measure and Manage; Article 9 of the EU AI Act requires a documented and regularly reviewed risk-management system for high-risk AI systems. A specific legal duty cannot be inferred from a tool label alone.

Under the EU AI Act’s terminology, an AI system is a machine-based system designed to operate with varying levels of autonomy, which may exhibit adaptiveness after deployment and infers from its inputs how to generate outputs such as predictions, content, recommendations or decisions that can influence physical or virtual environments. Legal classification depends on the system and its intended purpose. Not every automated process is therefore automatically an AI system for every regulatory purpose.

Cycle Time is the elapsed time between the defined start and defined finish of a completed work item. It includes all time within those boundaries, not only active work but also waiting and blockage. Individual values should be analysed as a distribution because averages can conceal variation and outliers.

Governance is the human-based system by which an organisation is directed, overseen and held accountable for its purpose. In operational practice, it includes authority limits, decision-making roles and rules, degrees of autonomy, assurance, reporting structures and accountabilities. Governance sets direction and boundaries; management organises work within that frame.

Human oversight means that appropriate natural persons can effectively oversee an AI system in its intended context. They need competence, authority, information and technical means to understand limitations and outputs, detect malfunction or misuse, intervene appropriately and stop the system where necessary. Article 14 of the EU AI Act specifies this requirement for high-risk AI systems; it does not create an identical obligation for every AI system.

Objectives and Key Results, or OKRs, combine a qualitative, directional Objective with a small set of measurable Key Results. The Objective describes the desired state; the Key Results provide testable evidence of progress towards it. OKR is a goal-setting and learning system, not a complete strategy process or a shared task list.

An operating model describes how an organisation executes its strategy and delivers value in day-to-day work. It connects capabilities, activities and processes, roles and structures, governance, data, technology, locations and performance management. It is therefore broader than an organisation chart or a process map.

An outcome is a short- or medium-term change that follows outputs, such as a change in behaviour, use, capability or a business condition. Unlike an output, it describes what changed for relevant stakeholders or within the system, not what was produced. An observed outcome is not automatically proof that one intervention caused it.

An output is the immediate product, service or work result produced by an activity. Examples include a released service, a trained team or an implemented capability. Outputs are usually more directly controllable by a team than outcomes, but they do not on their own demonstrate value or a desired change.

A product is a vehicle for delivering value. It has a recognisable boundary, known stakeholders, and clearly defined users or customers. A product may be a service, a physical offering or something more abstract, and it evolves over its lifecycle; it is not automatically the same as a time-limited project.

Product delivery translates prioritised product decisions into usable, quality-assured increments and makes them available to the intended users. It includes more than implementation: integration, testing, release, operation and feedback from real use also matter. Strong delivery creates value early and continuously without sacrificing quality or learning.

Product discovery is evidence-based work through which a product team investigates problems, user needs, context, risks and solution assumptions before and during significant delivery decisions. It uses research, data, prototypes and experiments to decide what to pursue, adapt or stop. Discovery may also show that no new product feature should be built.

A Product Goal describes a future state of the product that serves as a longer-term planning target for the product team. In the Scrum Guide, it is the commitment for the Product Backlog; the rest of the backlog emerges to define what might fulfil it. A Product Goal creates focus but does not prescribe the exact solution or guarantee an outcome.

Strategy execution is the systematic translation of strategic choices into prioritised initiatives, resources, accountabilities and operational work. It includes a testable target state, selection and funding of relevant work, clear governance, and regular performance and outcome feedback. It is therefore more than delivering a fixed plan: new evidence may require changes to initiatives or to the strategy itself.

A target operating model, or TOM, describes the intended future state of key organisational features and the route towards implementation. It makes visible how the organisation is expected to deliver services, make decisions, and connect capabilities, people, processes, data and technology. A TOM is a contextual design choice, not a universal ideal model.

Throughput, sometimes called delivery rate, is the exact number of work items finished per unit of time. It describes completion within the chosen system, not revenue, value or estimated points. Comparisons are meaningful only when the work-item type and the relevant boundaries are sufficiently stable.

Work in Progress, or WIP, is the number of work items that have started but not finished according to the system’s explicit definitions. WIP makes parallel work and committed attention visible. Its meaning depends on consistent start and finish points and a clear definition of the work items counted.

Work Item Age is the elapsed time from the defined start of an unfinished work item to the current measurement point. It shows how long current work has already been in the system and continues to increase until the item is finished. It is a snapshot of open work, not a completed Cycle Time.