Glossary · Executives & Directors

#

10-20-70 rule: BCG's description of how AI leaders spend resources: about 10% on algorithms, 20% on technology and data, 70% on people and processes. The AI Transformation Challenge
90-day AI plan: A contract for one quarter: a few measurable outcomes, one owner each, the dependencies they need and three decision dates. Building the 90-Day AI Plan
90-day replacement test: Asking whether you could replace a provider within 90 days, and what would block you, without necessarily intending to migrate. Model and Third-Party Risk

A

Abstention: The system declines, asks a clarifying question or escalates when evidence is missing, unreadable or unclear. Accuracy, Hallucination and Reliability
Acceptance threshold: The level of each relevant error the organization can tolerate for a use, written down before testing begins. AI Evaluation and Approval Gates
Accountable: The one person who answers for the outcome and holds the authority to decide. It cannot pass to a vendor or a machine. AI Decision Rights and Accountability
Activity metric: A count of AI use - licenses, users, prompts or pilots - that shows adoption, not a business result. Measuring AI Business Value
Adaptation: Fitting a pretrained model to a use through instructions, your documents, tools or fine-tuning. Foundation Models
ADMT: Automated decision-making technology. California rules require notice, opt-out and access for significant decisions from 2027. Privacy and Confidential Data
Adoption leak: The stage - aware, tried, weekly use, embedded, changing results - where people drop out. Each stage needs its own remedy. AI Change Management, KPIs and Operating Rhythm
Adoption theater: Visible, growing AI activity with no business number moving; everyone is busy with AI and the business has not noticed. Measuring AI Business Value
Agent assist: AI works beside a human service agent - summarizing, suggesting replies and finding policy - while the agent reviews and stays responsible. AI in Customer Service
Agent loop: Plan, act, observe, decide - repeated until the goal is met or a stopping condition ends it. From AI Assistants to AI Agents
Agent mandate: The written delegation for an agent - its outcome, systems, actions it may take alone, handoff rules and business owner. AI Agents and Intelligent Workflows
Agent washing: Rebranding assistants, chatbots or scripted automation as agents without real planning, tool use or action. Multimodal and Agentic AI
Agentic AI: An AI system that pursues a goal in a loop - planning, calling tools and acting - without a new human instruction at each step. Agentic AI and Autonomous Actions
AI adoption: The degree to which people, teams and processes actually change how they work using AI capabilities. AI Strategy vs AI Adoption
AI adoption curve: The path from owning AI tools to redesigned work, as AI moves from licenses to people, into processes and into how the business competes. The AI Adoption Curve
AI advantage: The gain AI adds on top of the best credible alternative, such as a rule, a checklist, a redesign or a hire. Finding Strategic AI Opportunities
AI advantage stack: Six layers from foundation models to customer experience; the higher the layer, the harder it is to copy. The AI Competitive Advantage
AI agent: A system that pursues a goal through a sequence of actions, using tools, information and permissions it has been given. From Copilots to AI Agents · also in From AI Assistants to AI Agents
AI ambition: The strategic choice of how much AI should change the organization's products, processes, workforce and competitive position. Defining AI Ambition
AI awareness: Knowing that AI matters to your industry and organization, without yet knowing what it can reliably do or what to change. From AI Awareness to AI Leadership
AI capability: The organization's ability to perform an AI activity well and repeatedly, through people, method, tools and rules. AI Talent and Capability Strategy
AI change management: The deliberate work of helping people adopt, and keep using, a new way of working that AI makes possible. AI Change Management, KPIs and Operating Rhythm
AI-down test: Asking whether a critical process can continue without its AI, for how long and at what capacity. Operational and Workforce Risk
AI-enabled professional: Someone who uses AI well, safely and critically in their own role without being a technical specialist. The AI Talent Gap
AI-enabled revenue: Revenue from an existing business that performs better because of AI, as opposed to customers paying for an AI product. Where AI Creates Revenue
AI-enhanced product: An existing product with a smarter feature; it still delivers its core value without the AI. AI-Native Products and Business Models
AI follower: An organization with the same AI that starts with tools, spreads effort across disconnected pilots and counts activity instead of outcomes. AI Leaders vs AI Followers
AI governance: The system of decision rights, accountability, policies, controls and oversight that keeps AI use responsible and aligned with business objectives. What Is AI Governance? · also in AI Policy vs AI Governance
AI Governance Committee: A cross-functional body with written authority to set AI direction, take material decisions, accept material risk and handle escalations. AI Governance Committee and AI Governance Office
AI Governance Office: The operating function that runs AI governance day to day - intake, routing, inventory, standards, decision support and follow-up. AI Governance Committee and AI Governance Office
AI hazard: An event involving an AI system that could plausibly lead to an AI incident; a near miss worth learning from. Human Oversight and AI Incidents
AI incident: An event in which developing, using or a malfunction of an AI system directly or indirectly leads to harm. Human Oversight and AI Incidents
AI inventory: A living record of every AI use, with its owner, purpose, data, provider, users, actions, legal and internal tier, and controls. AI Inventory and Risk Classification
AI islands: Separate, overlapping AI platforms built by different units that cannot share data, controls or reuse. Centralized vs Federated AI
AI leader: An organization whose habits - problem choice, ownership, redesign and proof - turn widely available AI into measured changes in how work is done. AI Leaders vs AI Followers
AI leadership: Changed behavior: sponsoring AI work, asking for evidence, redesigning processes and owning the decisions AI touches. From AI Awareness to AI Leadership
AI literacy: Understanding what AI can and cannot reliably do well enough to reason about its use, risks and requirements. From AI Awareness to AI Leadership
AI maturity: An organization's ability to identify, build, deploy, operate, govern, measure and scale AI value, repeatedly. The AI Maturity Model
AI-native product: A product designed around work the AI itself completes; remove the AI and nothing worth buying remains. AI-Native Products and Business Models
AI-native workflow: A process redesigned around AI capabilities, so steps that only moved information between people disappear and people handle the exceptions. Autonomous Workflows and AI-Native Organizations
AI operating model: The system of people, decision rights, funding, processes, technology and governance that lets an enterprise execute its AI strategy repeatedly and at scale. AI Operating Model
AI opportunity: A meaningful business improvement that AI may help deliver, stated in business terms before any tool is named. Finding Strategic AI Opportunities
AI-orchestrated workflow: A workflow in which AI coordinates several steps across several systems: gathering context, checking policy, preparing actions and routing work. Autonomous Workflows and AI-Native Organizations
AI platform: Reusable capabilities that let many teams build, test, secure, run and monitor AI solutions without rebuilding the foundations. AI Platform Strategy
AI policy: A formal statement of the organization's requirements and boundaries for using, building and buying AI. AI Policy vs AI Governance
AI portfolio: All AI experiments, products, automations and shared capabilities the enterprise funds, managed together against one budget and one pool of people. Prioritizing the AI Portfolio
AI risk: The potential business impact when an AI-enabled system produces, amplifies or acts on an incorrect, unsafe, unauthorized or inappropriate outcome. The AI Risk Landscape
AI strategy: The choices about why AI matters, where it applies, what comes first, who owns decisions and how much to invest. AI Strategy vs AI Adoption
AI system (OECD and EU): A machine-based system that infers from its input how to generate outputs such as predictions, content, recommendations or decisions. What Exactly Is Artificial Intelligence?
AI transformation: Changing how an organization works so that an AI capability produces repeatable business results, not just a successful demonstration. The AI Transformation Challenge
AI translator: A person with enough domain knowledge and AI literacy to turn a business problem into sound AI work, or to say no. The AI Talent Gap
AI value: The measurable business benefit AI causes by changing work, compared with what would have happened without it, net of cost. What Does AI Value Actually Mean?
AI value scorecard: A one-page executive view of one AI initiative that ends in a decision: scale, adjust, continue, pause or stop. AI Value Scorecard
AI washing: Claiming that a product or service uses AI, or uses it more capably, than it actually does. What Exactly Is Artificial Intelligence? · also in Module 03 Synthesis — Choosing Value Over Hype
Allocation rule: One written method for charging shared platform costs to AI systems, applied to every case so nothing is hidden or counted twice. Understanding AI Total Cost of Ownership
Ambient AI: AI that watches or listens continuously and acts when what it perceives matches a trigger, rather than waiting to be asked. Multimodal and Agentic AI
Ambition gap: The distance between the desired role of AI and the organization's current capability, which the strategy must explain how to close. Defining AI Ambition
Ambition statement: One sentence naming where AI changes the work, how deeply, by when, at what cost, and at least one boundary. Defining AI Ambition
Anchor problem: One real business problem, not an AI project, that a leader carries through every module as a learning reference. Your AI Leadership Starting Point
Anomaly: Behavior that deviates from a learned baseline. A signal that needs context, not proof of an attack. AI in Cybersecurity
Approval gate: An owned decision, based on evidence and known risk, on whether a system may move to the next stage and under what conditions. AI Evaluation and Approval Gates
Arrow register: A list of the assumptions between capability and outcome, each with the person who controls it and the cheapest early check. From AI Capability to Business Outcome
Artificial general intelligence: Hypothetical AI broadly competent across unrelated tasks. There is no agreed definition or test, so treat the term as a claim. The Major Types of AI
Artificial intelligence: The capability of a computer system to perform tasks that normally require human thinking, such as recognizing, predicting, generating or deciding within limits. What Exactly Is Artificial Intelligence?
Asymmetric error cost: When forecasting too low and too high cost different amounts, so the plan should lean toward the cheaper mistake. AI in Prediction, Forecasting and Optimization
Attention: The Transformer mechanism that lets each token weigh the earlier tokens in its context, so context steers the prediction. Large Language Models
Augmentation: AI raises the speed or quality of a person's work while that person stays in control and owns the outcome. AI Is Changing Everything
Authoritative source: The one system or document the organization has declared allowed to answer a given kind of question. AI in Knowledge Management
Automate versus augment: Whether AI replaces a task or strengthens the person doing it. The same exposure can shrink hiring or raise performance. AI and the Future of Work
Automated employment decision tool: Software that substantially assists hiring or promotion decisions. In New York City it needs a yearly independent bias audit and candidate notice. AI in Human Resources
Automation: A machine performs a repetitive task end to end, usually to cut cost, time or errors. AI Is Changing Everything
Automation bias: Relying on an automated aid in place of vigilant checking, so people miss what it misses and follow it when it is wrong. Human Oversight and AI Incidents
Automation paradox: Automation makes normal days easier and failures harder, because the people who must recover practice less. Operational and Workforce Risk
Average error: How far off a forecast is on a typical day, often as a percentage (MAPE). It can hide the rare days that matter most. AI in Prediction, Forecasting and Optimization

B

Backpropagation: The method that works backward from an error to calculate how each parameter should change to reduce it. Deep Learning — The Executive Mental Model
Baseline: The measured starting point, taken before launch and measured exactly as the result will be, against which change is judged. Baselines, Metrics and Measurement
Benefit owner: The named business leader accountable for a benefit line; finance validates the money. AI ROI and Value Realization
Benefit variance: The gap between planned and realized benefit, split by cause: adoption, benefit per use or conversion. AI ROI and Value Realization
Bias: A systematic pattern that produces consistently different outcomes for particular people or groups, with or without intent. Bias and Fairness
Big bet: A large commitment that pays off handsomely in some futures and loses heavily in others. Preparing for AI Uncertainty and Strategic Change
Blast radius: Everything a manipulated AI system could access, change, send or trigger before someone stops it. Security and AI Attacks · also in Agentic AI and Autonomous Actions
Borrowed conditions: Temporary advantages a pilot enjoys, such as clean data, expert reviewers and engineers on call, that production will not provide. From AI Pilot to Production to Scale
Bottom-up adoption pressure: Employees adopt a technology before the organization approves it, so leaders must see and bound use, not only introduce it. The Speed of AI Adoption
Bounded experiment: A visible trial with a few written rules, a measure and a date to keep, change or stop it. The Speed of AI Adoption
Break-even utilization: The owned or committed price per available hour divided by the rented price; below it, renting is cheaper. Model, Compute and Infrastructure Costs
Break-even volume: Fixed cost divided by contribution per unit: the volume at which total value covers total cost. AI Unit Economics and Economics at Scale · also in Module 08 Synthesis — Making AI Economically Sustainable
Brussels effect: Firms adopting EU rules worldwide because one global standard is cheaper than several (Anu Bradford, 2020). AI, Regulation, Geopolitics and Global Competition
Budget-line test: A cost saving is real only when you can name the budget line that falls, by how much, after AI's own costs. Where AI Reduces Cost
Build: Developing a significant AI capability inside the organization, owning its behavior, data and roadmap. Build vs Buy vs Partner
Build-to-buy seam: The gap that opens when governance rules bind AI built in-house but not AI bought from vendors or partners. Module 07 Synthesis — Governing AI at Scale
Bullwhip effect: Order swings that grow at each tier upstream because each tier reacts to the orders it receives, not to end consumption. AI in Operations and Supply Chain
Business as usual: The outcome expected if current arrangements continue and the proposal is not implemented; the benchmark every option is compared with. Building the AI Business Case
Business email compromise: Fraud that impersonates executives or suppliers to redirect payments. Cost 3.05 billion dollars in FBI complaints in 2025. AI in Cybersecurity
Business-first AI strategy: Choosing AI investments by starting from business goals, competitive position and constraints, rather than from available AI capabilities. Start With Business Strategy
Business metric: A number a business owner already manages, such as cycle time, cost to serve, quality or risk, that the initiative should move. Measuring AI Business Value
Business owner: The leader with authority over the process who is accountable for the AI outcome, as distinct from the team that builds the AI. The AI Transformation Challenge
Business productivity: Valuable output produced relative to the resources used: people's time, technology, capital and outside services. AI and Workforce Productivity
Buy: Acquiring a mature capability from the market; it still needs decisions on data, security, integration and governance. Build vs Buy vs Partner

C

Calibration: How well a system's stated confidence matches how often it is actually right. Accuracy, Hallucination and Reliability
Canary: A partial, time-limited deployment evaluated to decide whether to continue the rollout. From AI Pilot to Production to Scale
Capability: Something the technology can do, such as summarize, classify, predict or generate. On its own it says nothing about value. From AI Capability to Business Outcome · also in The Enterprise AI Use-Case Landscape
Capability blueprint: A short list, written as verbs, of what the organization must be able to do for its chosen AI ambition. AI Talent and Capability Strategy
Capability dials: The four things still changing fast: reliability, cost per task, autonomy and reach into the physical world. Where AI Is Going Next
Capability profile: What a particular system does reliably, task by task - strong at some tasks, weak or blind at others. What Exactly Is Artificial Intelligence?
Capability radar: A list sorting each tracked capability into Adopt, Trial, Watch or Park, moved only by evidence. Where AI Is Going Next
Capability-to-value chain: Capability, adoption, behavior change, outcome, value. Each link is necessary; none is sufficient on its own. What Does AI Value Actually Mean?
Capability versus reliability: What a system can do on its best day versus how dependably it does it; fluent output can still be wrong. Why Executives Need AI Literacy
Capacity creation: Time and attention AI frees. It becomes value only when management assigns it to output, quality, growth or lower cost. AI and Workforce Productivity
Capacity plan: A decision, made before scaling, on where released capacity goes, who owns the benefit and how it is measured. Productivity vs Realized Capacity
Catastrophic forgetting: A neural network losing skills it had when it is trained on new material, which is why every retrained version needs testing. Training vs Inference
Centralized AI: One enterprise function owns most AI capability and decisions; business units consume what it provides. Centralized vs Federated AI
Champion: Someone who promotes and encourages a change. Valuable, but without the powers to fund, stop or settle trade-offs. AI Transformation Governance and Executive Sponsorship
Change load: The number of planned changes a team absorbs in a period, AI and otherwise. A limit to manage, like engineering capacity. AI Change Management, KPIs and Operating Rhythm
Chargeback: Moving each team's share of spend into its own budget, once the attribution is trusted. Cost Optimization and AI FinOps
Cheat sheet: One page per level that gathers every chapter's card: one idea, key points, terms, frameworks, red flags and interview questions. Welcome and How This Program Works
Claim walkthrough: Testing a value claim step by step through definition, scope, adoption, capture and cost to find where it loses most. Module 03 Synthesis — Choosing Value Over Hype
Claims library: An owned, current list of approved product capabilities, prices, statistics and comparisons that AI-generated content may draw on. AI in Sales and Marketing
Closed-loop management system: Kaplan and Norton's cycle in which execution evidence is monitored and used to test and adapt the strategy itself. Module 09 Synthesis — From Strategy to Execution
Coherence: Choices that reinforce each other, so carrying out one makes the others easier rather than harder. Module 04 Synthesis — The Executive AI Strategy
Commitment gradient: The falling firmness of a roadmap over time: a quarterly contract, a one-year commitment, a multi-year direction and open options. Building the Multi-Year AI Roadmap
Compensating control: An extra safeguard, such as full human review, that holds down the added risk while an exception is open. AI Policies, Standards, Monitoring and Audit
Competitive advantage: Producing at lower cost than rivals, or delivering more perceived value, or a mix of the two (Rumelt). AI and Competitive Advantage
Complements: The processes, skills, data and organizational changes that a general-purpose technology needs before it pays off. The Four Industrial Revolutions
Computer-using agent: An AI agent that operates software by reading screenshots and clicking and typing, as a person would. Multimodal and Agentic AI
Concentration risk: Many business processes depending on one provider, region or cloud, so a single failure or change reaches all of them at once. Model and Third-Party Risk
Confidence mark: A low, medium or high rating beside each figure, judged from the amount, quality and agreement of the evidence behind it. AI Value Scorecard
Connector: A deliberately exposed interface that gives an agent a defined set of actions and data, with permissions and logs. Multimodal and Agentic AI
Constraint: The step or resource that limits a system's output; improving anything else does not raise the result. Start With Business Strategy
Constraint layer: The capability layer whose weakness currently limits what the organization can achieve with AI. The AI Talent Gap
Contestability: A real route for people affected by a consequential AI outcome to question it and have a person review it. AI Governance Principles
Context engineering: Designing everything a model receives when it runs - instructions, request, knowledge, data, tool results and memory - so it works well and safely. Prompting and Context Engineering — Executive Mental Model
Context window: The limit on how many tokens a model can take into account in a single request. Tokens, Context and Embeddings
Contribution per unit: Value of one unit minus the variable cost of delivering it, including expected review and failures; what remains to cover fixed cost. AI Unit Economics and Economics at Scale
Controlled retirement: Switching an AI system off deliberately - access removed, records archived, data handled, contracts ended, users told. AI Lifecycle Governance
Convergence: Several technologies maturing at roughly the same time, so that each one makes the others useful. Why AI, Why Now?
Conway's law: Organizations produce designs that copy their own communication structures; processes inherit the boundaries between departments. Autonomous Workflows and AI-Native Organizations
Copilot: An AI assistant that drafts, suggests or summarizes while a person stays in control and takes the action. From Copilots to AI Agents
Copy test: Asking what would still be hard to copy if a competitor got exactly your AI model tomorrow. The AI Competitive Advantage
Cost avoidance: Future spend lower than forecast; real value that never shows as a lower bill and must be reported separately. Where AI Reduces Cost
Cost of a first try: What a person must buy, install, learn or ask permission for before trying a technology; for generative AI it is close to zero. The Speed of AI Adoption
Cost of an outcome: The whole cost of delivering one business result, such as a correct listing, including people, data, checks and change. The Economics of AI
Cost of delay: The value lost for each month a useful capability is not yet live. Build vs Buy Economics and Investment Decisions
Cost of quality: Spending on prevention, appraisal, internal failure and external failure; errors found by customers cost the most. Where AI Reduces Cost
Cost per resolved outcome: Everything one finished piece of work costs, including retries and human review, not just one model call. Module 08 Synthesis — Making AI Economically Sustainable
Cost per successful task: Total cost of a physical deployment, including maintenance, supervision and idle time, divided by tasks completed correctly. AI + Robotics
Cost-to-serve: The complete cost of delivering one finished business outcome, including model, data, integration, operations and people. Data, Integration and Operational Costs
Counterfactual: What would have happened without the AI initiative; estimated with a baseline, comparison team or staggered rollout. What Does AI Value Actually Mean? · also in Baselines, Metrics and Measurement
Customer effort: The work a customer must do to get what they need - waiting, repeating, being transferred, switching channel. AI and Customer Experience

D

Data cascade: A data problem, often small and unnoticed at first, that compounds into failures further down an AI system. Data, Models and Compute
Data minimization: Using only the data a task needs, for a defined purpose, with the least access and the shortest retention that still works. Privacy and Confidential Data
Data moat: A lead that widens with use because more data keeps improving the product in ways a rival cannot buy. Proprietary Data, AI Moats and Differentiation
Data poisoning: Planting tainted material in the data a model learns from so it misbehaves later, often long after the plant. Security and AI Attacks
Data product: A data set run like a product - a named business owner, defined users, quality promises, documentation and a lifecycle. Data Strategy for AI
Data scale effect: More users create more data that improves the product for everyone, but only until the learning runs into diminishing returns. Proprietary Data, AI Moats and Differentiation
Data strategy for AI: The data requirements AI adds - knowledge, context, evaluation data, access rules - layered on the existing enterprise data strategy. Data Strategy for AI
Decision clock: How often, and how fast, a decision can actually change - minutes for a machine, weeks for a production plan. AI in Operations and Supply Chain
Decision control: Approving a decision and checking how it turned out, kept apart from proposing and running it. AI Decision Rights and Accountability
Decision log: The record of each material decision - rationale, risk accepted, conditions, owner and review date. AI Governance Committee and AI Governance Office
Decision milestone: A dated point where a named owner chooses between defined options on defined evidence, such as go or no-go, or fund or stop. Building the Multi-Year AI Roadmap
Decision right: The explicit authority to make a specific decision, with known limits and a known route for escalation. AI Decision Rights and Accountability
Decision rights: Clear answers to who may propose, build, approve, deploy, change and stop an AI system, and who is accountable for its outcome. What Is AI Governance?
Decision rule: A rule agreed before results arrive that says which incremental result, with which guardrails, leads to scale, iterate or stop. Baselines, Metrics and Measurement · also in AI Value Scorecard
Deep learning: Machine learning that uses neural networks with many layers to learn useful representations directly from data. Deep Learning — The Executive Mental Model
Deflection: The share of contacts that never reach a person. It counts ended conversations, not solved problems. AI in Customer Service
Deployer: Under the EU AI Act, an organization that uses an AI system under its own authority. It has duties separate from the provider's. What Is AI Governance?
Difference-in-differences: The change in the group that got AI minus the change in a comparable group that did not, over the same period. Baselines, Metrics and Measurement
Differentiation test: Ask whether we would still have an advantage if rivals had this capability tomorrow. If yes, it is not our edge. Build vs Buy vs Partner
Diffusion model: A generator, common for images and video, that starts from random noise and removes it step by step toward the prompt. How Generative AI Works
Digital Omnibus on AI: The 2026 EU regulation amending the AI Act; it moved high-risk dates to December 2027 and August 2028. AI, Regulation, Geopolitics and Global Competition
Diseconomies of scale: Cost per unit rising with volume, through harder cases, rising review shares, step costs or lower value per unit. AI Unit Economics and Economics at Scale
Distillation: Using a large model to train a smaller, cheaper one that keeps the behavior a specific task needs. Data, Models and Compute
Distinctness test: A second benefit counts only if it is a different resource, survives without the first and lands on a different account line. Total Business Impact
Diverge then converge: AI widens the set of options quickly; a small group of accountable people selects and tests the few worth building. AI in Product Development
Doom loop: A customer trapped in repetitive, unhelpful automated replies with no route to a human. The term appears in US regulator findings. AI in Customer Service
DORA delivery measures: Change lead time, deployment frequency, recovery time, change fail rate and rework rate - whether change reaches users fast and holds up. AI in Software Engineering
Double counting: Claiming the same economic benefit more than once, through alternative uses, relabeling or counting an asset and its income. Total Business Impact
DPIA: Data protection impact assessment under GDPR Article 35, required before processing personal data that is likely to be high risk. Privacy and Confidential Data · also in Why Responsible AI Matters
Draft then account: AI generates analysis or commentary; finance validates it and stays accountable before anything is issued. AI in Finance
Drift: A change after deployment in the inputs, or in the link between inputs and outcomes, that makes learned patterns less accurate. How Machine Learning Learns

E

Economic drift: Movement in cost, volume or value after approval that makes the business case stop describing the system. Module 08 Synthesis — Making AI Economically Sustainable
Economic sustainability: Each outcome earns more than it costs, the whole life is paid at real volume, and both stay true over time. Module 08 Synthesis — Making AI Economically Sustainable
Effective challenge: Critical review by objective, informed people with enough standing to get a model changed; a second-line job, not ownership. AI Roles, Ownership and Accountability
Effective cost per successful outcome: Model cost plus the cost of handling failures, divided by the outcomes actually delivered. Model, Compute and Infrastructure Costs
Embedded AI: AI built into software bought for another purpose - forecasting, ranking, routing or fraud scoring - often without the AI label. AI Is Already Inside Your Organization
Embedding: A learned list of numbers that places content by meaning, so similar items sit close together and can be compared. Tokens, Context and Embeddings
Enabler: A shared capability, such as clean master data, that several initiatives need and that has little standalone return. Prioritizing the AI Portfolio
Enterprise AI readiness: The organization's ability to turn a specific AI ambition into repeatable, governed, economically sustainable work. Assessing Enterprise AI Readiness
Enterprise AI strategy: A coordinated set of choices about where AI matters, what the organization builds and funds, and what it will not do. What Is an Enterprise AI Strategy?
Escalation: A designed handover to a human, with full context, when a request is outside the AI's limits or confidence. AI in Customer Service
Escalation of commitment: The tendency to invest more in a failing course of action one is personally responsible for. Prioritizing the AI Portfolio
Escalation trigger: A condition agreed in advance that moves a delegated decision up a level automatically. AI Decision Rights and Accountability · also in AI Governance Committee and AI Governance Office
Evaluation: Structured testing of whether an AI system meets defined requirements, and what limits, risks and behavior remain. AI Evaluation and Approval Gates
Evaluation data: Representative cases, expected outcomes, edge cases and known failures used to show that an AI system works. Data Strategy for AI
Evaluation set: A fixed, representative set of real questions with expected answers, run against every prompt or model change before release. Prompting and Context Engineering — Executive Mental Model · also in The Evolution of AI Models
Evidence ladder: Model prediction, then customer statements, then behavior in a test, then behavior at scale. Each rung is stronger evidence. AI in Product Development
Evidence question: A request that names a claim, a baseline, a threshold, a guardrail and the decision the result will settle. From AI Awareness to AI Leadership
Exception cost: Cases sent to a person, times time per case, times the loaded hourly rate, every year. Data, Integration and Operational Costs
Exception path: The designed, staffed and owned route for cases that do not fit the normal path, including escalation and stop rules. Autonomous Workflows and AI-Native Organizations
Excessive agency: OWASP's name for a system with more functionality, permissions or autonomy than its job needs. Agentic AI and Autonomous Actions
Executive AI literacy: Enough understanding of AI to judge its strengths and limits, ask the questions that expose value and risk, and own the decision. Why Executives Need AI Literacy
Executive sponsor: A senior leader who owns an AI initiative's business outcome, can fund it into production and has the authority to stop it. AI Leaders vs AI Followers · also in AI Transformation Governance and Executive Sponsorship
Exit strategy: A plan for leaving a vendor or partner - data portability, migration effort, alternative suppliers and contract terms. Build vs Buy vs Partner
Expected downtime cost: The cost of one hour down multiplied by the hours you expect to lose; the yardstick for resilience spend. Data, Integration and Operational Costs
Expected value: Potential value after discounting for adoption and for how much freed capacity is realized; the forecast a case should present. Building the AI Business Case
Experience-to-value chain: Better interaction, then changed customer behavior, then a business outcome. Value appears only at the last link. AI and Customer Experience
Explain test: Rate your confidence, explain the topic in three plain steps, point to a decision where you used it, then re-rate. Your AI Leadership Starting Point
Exploitation likelihood: The predicted chance that a published vulnerability will be used by attackers. Only about 6 percent ever are. AI in Cybersecurity

F

Fairness drift: A system that was fair at launch becomes unfair as data, populations, behavior or policy change. Bias and Fairness
False negative: A real attack the system fails to flag or quietly suppresses. The error that ends up in the incident report. AI in Cybersecurity
Federated AI: Business units own their AI teams, priorities and outcomes, with light enterprise coordination at most. Centralized vs Federated AI
Feedback loop: The time between doing the work and knowing whether it worked. A first win needs one measured in weeks or months, not years. Where Should We Start With AI?
Fine-tuning: Further training on your own examples; good for behavior, format and style, poor for adding new facts. Foundation Models · also in RAG and Enterprise Knowledge — Executive Mental Model
FinOps: An operating practice in which the teams that create technology cost can see it, own it and improve it, in a continuous loop. Cost Optimization and AI FinOps
First win: A first AI project chosen to deliver real value and to teach the organization: hard enough to matter, realistic enough to finish. Where Should We Start With AI?
Fitness for use: Data quality judged against a specific purpose, not in general. The same data can suit one use and fail another. Data Strategy for AI
Fixed-cost absorption: Fixed cost spread over more units as volume grows; low adoption leaves the same cost on fewer units. AI Unit Economics and Economics at Scale
Fixed workflow: Steps set in advance by code, with AI used inside single steps; cheaper and easier to test than an agent. From AI Assistants to AI Agents
Flip point: The volume, price or capability change at which the preferred sourcing option would change. Build vs Buy Economics and Investment Decisions
Flywheel: A loop where use creates feedback, feedback improves the offer and a better offer brings more use. AI and Competitive Advantage
Forecast range: The span reality is expected to fall in, such as 14 to 30 no-shows around a central 22, instead of one number treated as fact. AI in Prediction, Forecasting and Optimization
Foundation model: A general model trained once on broad data at scale and adapted to many tasks, such as drafting, summarizing and translating. Why AI, Why Now? · also in Foundation Models
Funding mechanism: The rules for who pays for shared platforms and who pays for business products, designed so money reinforces ownership and reuse. AI Operating Model

G

General-purpose technology: A technology that is pervasive, keeps improving and spawns complementary innovation, such as steam, electricity or computers. The Four Industrial Revolutions
Generalization: A model's ability to perform well on new cases that it did not see during training. How Machine Learning Learns
Generative AI: General-purpose AI that drafts, summarizes and transforms content in response to requests in ordinary language. Generative AI Changes the Game · also in The Major Types of AI, How Generative AI Works
Golden path: Spotify's term for the opinionated, supported route to build something on the platform, designed so the safe way is also the easy way. AI Platform Strategy
Governance as a service: One intake, visible status, stated service levels and reusable approved patterns that teams can use without inventing their own. The AI Governance Operating Model
Governance principle: A short, high-level expectation of how AI must behave, independent of any model or vendor, that guides real decisions. AI Governance Principles
Governance seam: A point where one part of governance hands work to the next, or where one unit's version of a rule meets another's. Module 07 Synthesis — Governing AI at Scale
Governance theater: Meetings, policies and dashboards that change no decision and no behavior. The AI Governance Operating Model
Grounded answer: An answer built from retrieved evidence, with sources a person can check against each claim. RAG and Enterprise Knowledge — Executive Mental Model
Groundedness: Whether an answer is supported by the source the system was supposed to use, such as current company policy. Accuracy, Hallucination and Reliability
Grounding: Making the AI answer only from approved, current sources and say so when it cannot find one. AI in Customer Service
Guardrail: A control on spending, from an alert to a named owner up to a hard limit that stops usage at a ceiling. Cost Optimization and AI FinOps
Guardrail metric: A measure that must not get worse, such as rework or stability, while the primary metric improves. Leading vs Lagging AI Metrics

H

Hallucination: Fluent model output that is false or unsupported, such as an invented fact, figure or reference. Large Language Models · also in Accuracy, Hallucination and Reliability
Handoff: A case the agent stops on and passes to a named person, with the request, what it checked and why it stopped. AI Agents and Intelligent Workflows · also in Module 09 Synthesis — From Strategy to Execution
Hidden pilot labor: Work done by hand during a pilot, such as checking outputs, that its budget omits and production must price or remove. Experimentation vs Production Economics
High-risk HR AI: Under EU AI Act Annex III, AI for recruiting, promotion, termination, task allocation or performance monitoring. Obligations apply from 2 Dec 2027. AI in Human Resources
Holdout group: Customers deliberately left out of a campaign or AI treatment so its effect on outcomes can be measured. AI in Sales and Marketing
Homogenization: Many applications built on the same few models, so they share strengths and also inherit the same flaws. Foundation Models
Hub and spoke: A hybrid in which a central hub owns governance, standards and talent strategy, and business-unit spokes own adoption. Centralized vs Federated AI
Human authorship: The US requirement that copyright protects what a person contributed - selection, arrangement, modification - not purely AI-generated material. Intellectual Property and Copyright
Human in the loop: AI extracts or recommends, a person approves, then the process runs. Used where errors are costly. AI in Document and Data Processing
Human on the loop: AI processes on its own while people monitor the flow and override when needed. Used for high-volume, lower-risk work. AI in Document and Data Processing · also in Human Oversight and AI Incidents
Hype: A claim in which a phrase such as everyone is doing it stands where a measured link in the value chain should be. Module 03 Synthesis — Choosing Value Over Hype

I

If-then plan: A commitment of the form 'when X happens, I will do Y', which makes follow-through more likely than a general intention. Your AI Leadership Starting Point
Illusion of explanatory depth: The tendency to feel you understand how something works far better than you can actually explain it. Your AI Leadership Starting Point
Imperfect imitability: Barney's term for resources rivals cannot easily copy, because of history, causal ambiguity or social complexity. The AI Competitive Advantage
Incremental revenue: Revenue that would not have happened without the AI change, estimated against a control group or other counterfactual. Where AI Creates Revenue
Independent assurance: Internal audit's objective judgment, separate from management, on whether controls are well designed and operating as intended. AI Policies, Standards, Monitoring and Audit
Indirect prompt injection: Hostile instructions planted in a document, email or web page that the AI later reads; the attacker never talks to it. Security and AI Attacks
Inference: Using a trained model to turn an input into an output. The parameters stay unchanged. Training vs Inference · also in Model, Compute and Infrastructure Costs
Informal AI: AI tools employees bring themselves - personal accounts, extensions, departmental subscriptions - outside approval. Often called shadow AI. AI Is Already Inside Your Organization
Informed trust: Trust matched to what a system can really do: people know its limits, its owner and how to challenge it. Why Responsible AI Matters
Inherent risk: The risk of a use case as designed, before any control is applied. Module 06 Synthesis — Understanding AI Risk
Input-pricing trap: Pricing by seats or hours, so revenue shrinks as the product lets customers do more with fewer of them. AI-Native Products and Business Models
Installation and deployment: Perez's two periods of a revolution: speculative build-out first, broad productive use later, often after a crash. The Four Industrial Revolutions
Integration upkeep: Connections times interface-affecting changes per year times the cost of absorbing one change. Data, Integration and Operational Costs
Isolating mechanism: Whatever stops rivals from closing an advantage, such as contracts, relationships, reputation, scale or tacit know-how. AI and Competitive Advantage

J

Jagged frontier: The uneven boundary of AI capability: similar-looking tasks can fall inside it, where AI helps, or outside it, where AI hurts. AI Is Changing Everything

K

Knightian uncertainty: An outcome whose odds cannot be estimated with confidence, unlike risk, where the odds can be measured. Preparing for AI Uncertainty and Strategic Change
Knowledge owner: The named person or team accountable for keeping a knowledge domain correct, current and appropriately visible. AI in Knowledge Management
Knowledge stickiness: The difficulty of moving a good practice from one part of an organization to another. AI in Knowledge Management

L

Label choice bias: Bias created by predicting a convenient stand-in, such as health cost, instead of the outcome that matters, such as health need. Bias and Fairness
Ladder of finance authority: Four rungs: read and analyze, draft and recommend, act within limits, approve and execute. Each rung up needs its own controls. AI in Finance
Lagging metric: A business result, such as unit cost, revenue or retention, that moves late and proves whether value arrived. Leading vs Lagging AI Metrics
Language as the interface: Reaching AI by stating the outcome you want in ordinary words, instead of learning code, query languages or specialist screens. Generative AI Changes the Game
Large language model: A large neural network trained on vast text and code to predict the next token from context; the base of most AI assistants. Large Language Models
Lead scoring: A model that ranks leads by resemblance to past conversions. It orders the seller's queue; it does not qualify the lead. AI in Sales and Marketing
Leader's loop: Watch thresholds, test on your own work, commit at the speed a decision can be undone, and review on a fixed date. Module 10 Synthesis — The Future AI Leader
Leading metric: An early reading, such as use on target work or override rate, that has a record of moving before the outcome. Leading vs Lagging AI Metrics
Leakage: Information unavailable in real use slipping into training or testing, so the test score overstates real performance. How Machine Learning Learns
Learned model: A system whose behavior comes from patterns in example data, not from rules written line by line; it is checked by testing. AI vs Automation
Least privilege: Running each program or agent with only the privileges its task requires, and nothing because it might be useful later. Agentic AI and Autonomous Actions
Legal tier: The category the EU AI Act assigns to a use - prohibited, high-risk, transparency or minimal - regardless of internal scoring. AI Inventory and Risk Classification
Lethal trifecta: Private data, untrusted content and external communication in one system; together they let an attacker steal data. Security and AI Attacks
Lifecycle governance: The policies, decisions, controls, reviews and accountability applied to an AI system across its whole life, from discovery to retirement. AI Lifecycle Governance
Load-bearing assumption: An assumption whose failure would force a real change to the plan; Dewar's assumption-based planning starts by listing them. Building the Multi-Year AI Roadmap
Lock-in: Dependence built from proprietary features, tuned prompts, provider-specific tools and embeddings that makes replacing a provider slow and costly. Model and Third-Party Risk
Long pole: A dependency on the critical path, often in another function, such as a security review or data agreement, that sets the earliest finish. Building the 90-Day AI Plan
Loss function: The measure of how wrong each prediction is; training adjusts the model to make it smaller. How Machine Learning Learns

M

Machine learning: A program that improves at a task, as judged by a measure, from experience, instead of following rules written for every case. AI vs Machine Learning
Material change: A change that alters an AI system's risk, impact, data, users, autonomy or decision consequences, and so requires reassessment. AI Lifecycle Governance
Material scope: The agreed boundary of which systems and processes AI governance applies to, best shown with in-scope and out-of-scope examples. Module 07 Synthesis — Governing AI at Scale
Materiality threshold: The size of variance or error above which finance must investigate and explain, set by the controller. AI in Finance
Maturity profile: Maturity rated for each business unit and dimension, with evidence, instead of one enterprise score. The AI Maturity Model
Metered cost: Cost that moves with each request, page of context, answer and human check, unlike a license that sits still. The Economics of AI
Minimum required readiness: Building the capabilities the current stage and the next stage need, rather than every foundation before starting. Assessing Enterprise AI Readiness
Mixed diagnosis: An honest assessment that names where an organization leads and where it follows, function by function, with one strength and one gap. AI Leaders vs AI Followers
Modality: A type or channel of information, such as text, images, audio, video or sensor data. Multimodal AI
Model access layer: One controlled doorway to several AI models that can route each task by cost, quality and data sensitivity. AI Platform Strategy
Model portfolio: A small set of models of different capability and cost, each matched to the workloads that need it. The Evolution of AI Models
Model serving: Running a trained model in production: accepting requests, scaling with demand and returning answers fast enough and cheaply enough. Training vs Inference
Moral crumple zone: Elish's term: blame for an automated system's failure lands on the nearest human operator, who had little control over it. AI Roles, Ownership and Accountability
Moravec's paradox: Skills easy for a small child, such as perception and movement, are hard for machines, while abstract reasoning is comparatively easy. AI + Robotics
Multimodal AI: AI that can process, combine or generate more than one modality, in its inputs, its outputs or both. Multimodal AI

N

Narrow AI: AI that performs well within a defined task or domain. Narrow describes breadth, not strength; some narrow systems beat every human. The Major Types of AI
Net AI value: Total benefit minus total cost of an AI capability, with both sides tested at expected scale. The Economics of AI
Net present value: Future net cash flows discounted at the required rate of return, minus the investment. AI ROI and Value Realization
Net workflow gain: The end-to-end time saved after checking, rework, hand-offs and waiting at the constraint are subtracted. Productivity vs Realized Capacity
Never-skilling: Juniors never build expertise because AI now does the routine work they used to learn on. Operational and Workforce Risk
No-regret move: An action that pays off in every plausible future, such as a test set built from your own work. Preparing for AI Uncertainty and Strategic Change
Not-yet list: The written list of deferred work, each item with its reason and the quarter in which it will be reconsidered. Building the 90-Day AI Plan

O

Official AI: AI tools the organization licensed as AI, with a contract, a named owner and a usage policy. AI Is Already Inside Your Organization
Offloaded judgment: People still do the task but stop thinking it through, accepting work that looks finished. Operational and Workforce Risk
One-page AI plan: Ambition, weakest link, three to five quarterly outcomes, gate decisions, roadmap changes and owners. A blank line shows a broken loop. Module 09 Synthesis — From Strategy to Execution
One-way door: A decision that is consequential and nearly irreversible, so it deserves a slow, senior process; most decisions are two-way. Preparing for AI Uncertainty and Strategic Change
Open-weight model: A model whose trained parameters are published, so anyone can download, run and adapt it. The Evolution of AI Models
Operating effectiveness: Whether a control actually works as designed in daily practice, as opposed to whether its design would meet the objective. AI Policies, Standards, Monitoring and Audit
Operating model: How governance decisions are made, where authority sits, how requests flow and when they escalate - beyond the reporting lines. The AI Governance Operating Model
Operating rhythm: The fixed calendar of weekly, monthly, quarterly and annual reviews in which evidence is read and decisions are taken. AI Change Management, KPIs and Operating Rhythm
Opportunity cost: The net value of the best other work a team could deliver instead of this build. Build vs Buy Economics and Investment Decisions
Optimism bias: The proven tendency for appraisals to be over-optimistic about costs, durations and benefits. Building the AI Business Case
Optimization: Searching for the best action against a written objective, within hard limits that must hold and soft limits that bend at a cost. AI in Prediction, Forecasting and Optimization
Optimization order: Remove work, reshape workflows, route by difficulty, cache and batch, then negotiate rates. Cost Optimization and AI FinOps
Option (strategic): A small commitment now that secures the right to act at scale later while limiting losses if the future turns bad. Preparing for AI Uncertainty and Strategic Change
Out-of-band verification: Confirming a request through a separate, trusted channel, such as calling back on a number already on file. AI in Cybersecurity
Outcome capture: Recording what actually happened after each decision, inside the workflow, so the next decision can improve. Proprietary Data, AI Moats and Differentiation
Outcome hypothesis: For whom, using which capability, changing which step of work, so that which outcome moves, while which guardrail holds, owned by whom. From AI Capability to Business Outcome
Outcome owner: The named business leader accountable for the result an AI use case promised, whoever owns the technology. AI Operating Model
Outcome pricing: Charging per defined result, such as a resolved case, rather than per user or per request. AI-Native Products and Business Models
Over-trust: Approving AI output because it sounds fluent and confident, without checks matched to the cost of an error. Why Executives Need AI Literacy
Overfitting: Fitting the training examples so closely that the model learns their noise and coincidences and performs poorly on new data. How Machine Learning Learns
Override design: The rules for where people may change an algorithm's inputs or outputs, for which cases, and how each override is scored. AI in Operations and Supply Chain
Oversight duty: A director's duty to make a good-faith effort to have reporting on mission-critical risks and to act on red flags. Why Executives Need AI Literacy

P

Package hallucination: A code model recommending a software library that does not exist, a name an attacker can register with malicious code. AI in Software Engineering
Paper program: US prosecutors' term for a compliance program that exists on paper but is not implemented, resourced or reviewed. AI Policy vs AI Governance
Parameter: One of the many adjustable numbers inside a model; training changes them so that predictions improve. Deep Learning — The Executive Mental Model
Parity: An investment every competitor can make. It protects the business but does not set it apart. AI and Competitive Advantage
Partner: Combining our domain knowledge and data with a specialist's skills to create a capability neither could easily make alone. Build vs Buy vs Partner
Pattern mix: The combination of create, understand, predict, decide and act that a problem needs, as opposed to the one a team already owns. The Enterprise AI Use-Case Landscape
Payback: The time it takes to recover the investment; it ignores the time value of money and flows after payback. AI ROI and Value Realization
People decision: Any output that shapes someone's job, pay or opportunity: hiring, promotion, pay, shifts, performance or exit. AI in Human Resources
Permission-aware retrieval: Search that applies the asking user's access rights before any content reaches the model. RAG and Enterprise Knowledge — Executive Mental Model · also in Privacy and Confidential Data
Physical AI: AI that senses and acts in the physical world through robots, vehicles, drones or other machines, running a continuous loop rather than giving one answer. AI + Robotics
Pilot bill: The cost of a small trial. It tests ideas and assumptions; multiplied by volume, it is not a forecast for production. The Economics of AI
Pilot purgatory: Many AI experiments, few systems in production, and new pilots that hide the lack of conversion. The AI Maturity Model
Pipeline: A chain of specialist models, such as speech to text then a language model, joined by a coordinating layer. Multimodal AI
Planning fallacy: The tendency to underestimate how long one's own work will take, even when asked for a worst-case estimate. Building the 90-Day AI Plan
Polanyi's paradox: We know more than we can tell: people recognize many patterns they cannot write down as complete rules. AI vs Automation
Post-training: Training after pretraining, using examples and human ratings, that teaches a model to follow instructions helpfully and safely. Large Language Models
Posture: A steady way of operating that keeps an organization able to benefit whichever way AI moves, instead of betting on one forecast. Module 10 Synthesis — The Future AI Leader
Practice testing: Recalling material instead of rereading it. It improves retention after a week, even though rereading feels more reassuring. Welcome and How This Program Works
Pre-mortem: Before committing, imagine the initiative has already failed and list why. Prospective hindsight surfaces more reasons than asking what might go wrong. Module 10 Synthesis — The Future AI Leader
Preemption: Federal law overriding state law. In US AI policy it is being sought by executive action, which does not by itself repeal state laws. AI, Regulation, Geopolitics and Global Competition
Premature scale: Rolling out widely before economics, local context, support and governance are proven, so costs rise and value falls short. From AI Pilot to Production to Scale
Pretraining: The costly first stage, in which a model learns general patterns from vast, broad data without hand-labeled examples. Foundation Models
Price of an answer: What the model charges for one response. Small, visible on the invoice, and falling. The Economics of AI
Primary benefit: The financial benefit the case stands on, measured where the money appears, compared with the operating plan. Total Business Impact
Principle tension: A case where two principles pull apart, such as transparency and security; an accountable owner decides and records why. AI Governance Principles
Problem of many hands: Thompson's term: when many people contribute to a decision, it becomes hard to hold any one of them responsible. AI Roles, Ownership and Accountability
Process carries the AI: The workflow itself depends on AI, so results no longer rely on individuals remembering to use a tool. The AI Adoption Curve
Product discovery: Deciding what to build and testing that customers value it and can use it, before committing to build. AI in Product Development
Production bar: The quality and availability a workflow needs, set by what one error or one hour down costs the business. Experimentation vs Production Economics
Production case rebuild: A new business case at the gate from full TCO, target-user adoption, volume scenarios, an agreed bar and an owner. Experimentation vs Production Economics
Productivity paradox: Wide adoption of a technology with little measured productivity gain, as Solow observed for computers in 1987. The Four Industrial Revolutions
Prompt: The instructions, questions, examples and material given to a model when it is used; it changes the input, not the model. Prompting and Context Engineering — Executive Mental Model
Prompt injection: Untrusted text that steers an AI system to act against its owner's intent, typed directly or hidden in content it reads. Security and AI Attacks
Proportionality: Control depth matches the impact, likelihood and exposure of a use case; the principles themselves stay the same. AI Governance Principles
Proportionate governance: Controls that rise with the impact of an AI use, so low-risk experiments move fast and high-impact uses get stronger review. What Is AI Governance?
Provenance: The traceable chain from a source document and page to the extracted value, its checks and the decision it fed. AI in Document and Data Processing
Provenance record: A short record of the tool, inputs, human contribution, review and release decision behind an AI-assisted asset. Intellectual Property and Copyright
Provider and deployer: EU AI Act roles: the provider develops or brands a system; the deployer uses it. What you do decides which you are. AI Roles, Ownership and Accountability
Proxy variable: A variable, such as postal code, college or first name, that correlates with a sensitive attribute and can carry its signal. Bias and Fairness

Q

Quality-adjusted productivity: Output change times quality change. Thirty percent more output at 20 percent lower quality is only about a 4 percent gain. AI and Workforce Productivity
Quantization: Storing a model's numbers at lower precision so it needs less memory and runs faster, with a quality trade-off to test. Data, Models and Compute
Quick win: An initiative with a clear problem, available data and limited process change that returns measurable value within months. Quick Wins, Strategic Bets and Transformation Initiatives
Quick-win trap: A portfolio of many small, separately built tools that never adds up to a shared capability or a future option. Quick Wins, Strategic Bets and Transformation Initiatives

R

Readiness record: For each dimension, the current score, the evidence, the level the ambition requires, and the action that closes the gap. Assessing Enterprise AI Readiness
Realization rate: Realized benefit divided by expected benefit for the same period. 630,000 realized against 1 million expected is 63 percent. AI Value Scorecard
Realized capacity: Freed time or resources that the organization actually converts into measured output, avoided cost or another business result. Productivity vs Realized Capacity
Realized ROI: Measured benefit minus actual full cost, divided by actual full cost; known only after the money is spent. AI ROI and Value Realization
Reasoning model: A language model trained, largely with reinforcement learning, to work through intermediate steps before answering; stronger on multi-step problems, slower and costlier. Large Language Models
Reassessment trigger: A written change or event - purpose, data, model, autonomy, users, threshold breach, incident - that reopens an approval. AI Evaluation and Approval Gates
Rebound effect: When a resource gets cheaper to use, total use can grow so much that total spending rises. Also called the Jevons paradox. The Economics of AI
Recoverable error: A mistake that is seen and fixed before it counts, such as a draft a specialist checks before it is sent. Where Should We Start With AI?
Regression to the mean: The tendency of unusually bad or good numbers to move back toward normal on their own. Baselines, Metrics and Measurement
Reinforcement learning: Learning by acting and receiving rewards or penalties; central to tuning assistants on human preferences and training reasoning models. AI vs Machine Learning
Relevant range: The band of activity within which a fixed or step cost stays flat. Forecasts are valid only inside it. Understanding AI Total Cost of Ownership
Repeat-run reliability: How often an agent succeeds every time the same task is run again, not just once. AI Agents and Intelligent Workflows
Representation: An internal description of the input that a network learns for itself, such as edges, textures or parts, rather than one a person specifies. Deep Learning — The Executive Mental Model
Residual risk: The risk that remains after controls - the risk a named owner decides to mitigate, transfer, avoid or accept. Module 06 Synthesis — Understanding AI Risk · also in AI Evaluation and Approval Gates
Responsible: The person or team that performs the work. One decision can have several. AI Decision Rights and Accountability
Responsible AI: Designing and operating AI so it creates value while protecting people, managing risk and keeping clear accountability. Why Responsible AI Matters
Retrieval-augmented generation (RAG): Retrieving relevant, permitted information and giving it to a model as context before it generates an answer. RAG and Enterprise Knowledge — Executive Mental Model
Revenue protected: Revenue kept because a customer who would have left stays; real value, reported separately from new revenue. Where AI Creates Revenue
Review date: The date on which a commitment is decided again: keep, change or stop. A planned stop is a result, not a failure. Module 10 Synthesis — The Future AI Leader
Risk-adjusted TCO: Total cost of ownership plus the expected cost of overruns, repricing and other likely surprises. Build vs Buy Economics and Investment Decisions
Risk-adjusted value: Value if delivered, times the probability of delivery, minus full cost. Each discount is applied once. Building the AI Business Case
Risk appetite: How much risk an organization chooses to carry for a given value, within the limits the law already sets. Module 06 Synthesis — Understanding AI Risk
Risk-based routing: Using a system's risk to decide which reviews, approvals and controls it needs, so routine cases never join the specialist queue. The AI Governance Operating Model
Risk classification: Assigning each AI use a tier that decides who reviews it, who approves it and how closely it is monitored. AI Inventory and Risk Classification
Risk multipliers: Reversibility, reach, detectability and speed - the four questions that make the same error trivial or serious. The AI Risk Landscape
Role skill target: The AI skill level a role family must reach, from literacy to specialist, set deliberately and funded. AI Talent and Capability Strategy
Role-specific AI literacy: Enough understanding to judge AI output and redesign the work in your own role. It does not mean learning to code. AI and the Future of Work
Role track: A curated list of 10 to 14 Level 1 chapters for one role, such as CEO, CFO, COO, risk and legal, or CTO and CIO. Welcome and How This Program Works
Router: Software in front of the models that sends each request to the cheapest model that clears its quality bar. The Evolution of AI Models
Rule-based automation: Software that executes steps people specified in advance, such as if-this-then-that rules, giving the same result for the same input. AI vs Automation
Rules versus access: Rules govern what you may do in a market; access governs whether you can obtain chips, compute, models or markets at all. AI, Regulation, Geopolitics and Global Competition
Runtime data: Information supplied at the moment of use, such as a request and current records, that the model never saw in training. Data, Models and Compute

S

Safety controller: A deterministic layer that permits, slows or stops a machine's movements regardless of what the AI layer proposes. AI + Robotics
Sampling: Choosing each next piece of output by a weighted draw among likely options, rather than always taking the single likeliest. How Generative AI Works
Say- and do-metrics: Say-metrics record opinions such as satisfaction scores; do-metrics record behavior such as returns, repeat contacts and renewals. AI and Customer Experience
Say versus do: The gap between what customers request or rate and what their behavior shows they need. AI in Product Development
Seam failure: A break between two parts that each look healthy, such as a funded pilot with no named production owner. Module 09 Synthesis — From Strategy to Execution
Segregation of duties: Separating who prepares, reviews, approves and executes a financial action, so no single actor, human or AI, controls all four. AI in Finance
Self-supervised learning: Learning from answers hidden in the data itself, such as the next word of a text; how large language models are pretrained. AI vs Machine Learning
Shadow AI: Employees using unapproved AI tools with company data, usually because the approved route is slower or missing. Privacy and Confidential Data
Showback: Reporting each team's share of spend to it without moving the cost into its budget. Cost Optimization and AI FinOps
Signal to authority: The handoff from a monitoring alert to a named owner who has the authority to pause the system. Module 07 Synthesis — Governing AI at Scale
Signpost: An observable early signal that a vulnerable assumption is failing, with a named person watching it and an agreed response. Building the Multi-Year AI Roadmap
Silent error: A task the agent completed wrongly without anyone noticing; found only later, or by sampling completed work. AI Agents and Intelligent Workflows
Silent run: Running a system on live data and logging its outputs without anyone acting on them, before launch. Also called shadow mode. From AI Pilot to Production to Scale
Simulated respondents: Language models answering surveys as if they were customers. Useful to pre-test a study, unreliable for new products or segments. AI in Product Development
Six components: Technology, data, people, process, governance and leadership - the parts that must hold together before AI creates business value. The AI Transformation Challenge
Socio-technical: NIST's term for AI risk arising from technology together with how, where and by whom it is used. The AI Risk Landscape
Solely automated decision: A decision made without meaningful human involvement. EU and California rules regulate these whether a rule or a model makes them. AI vs Automation · also in AI in Human Resources
Spaced practice: Reviewing material in short sessions spread over time, such as a two-minute card a week later, rather than in one sitting. Welcome and How This Program Works
Spread and depth: Spread counts who has access or uses AI. Depth asks whether the work itself now depends on it. The AI Adoption Curve
Standard: A specific, checkable bar that translates a policy requirement into something teams can meet. AI Policy vs AI Governance · also in AI Policies, Standards, Monitoring and Audit
State: The agent's running record of where the job stands, so it neither repeats nor skips steps. From AI Assistants to AI Agents
Step cost: A cost that stays flat within a range of activity, then jumps to a new level, such as one more reviewer or capacity block. Understanding AI Total Cost of Ownership
Stop rule: The result, written before a test starts, that would make the team halt or redesign the use case. AI Use-Case Discovery and Design · also in Prioritizing the AI Portfolio
Stop sentence: A line naming what the strategy will no longer fund. Its absence usually means no real choice was made. Module 04 Synthesis — The Executive AI Strategy
Straight-through processing: The share of items that pass from intake to posting with no human touch. AI in Document and Data Processing
Strategic bet: A high-uncertainty investment that buys the option, not the obligation, to invest more if evidence supports it. Quick Wins, Strategic Bets and Transformation Initiatives
Strategic refusal: An explicit decision about what the organization will not do with AI, so scarce talent, data and money go to the priorities. What Is an Enterprise AI Strategy?
Strategy chain: The linked choices from business problem to advantage: why, where, what, how and win. Module 04 Synthesis — The Executive AI Strategy
Strategy kernel: Rumelt's three parts of a good strategy: a diagnosis of the critical obstacle, a guiding policy and coherent actions. What Is an Enterprise AI Strategy?
Substantial modification: Under the EU Machinery Regulation, an unplanned physical or digital change affecting safety; whoever makes it is treated as the manufacturer. AI + Robotics
Supervised learning: Learning from past cases paired with known answers, such as loans and whether each defaulted, to predict answers for new cases. AI vs Machine Learning
Supporting indicator: A metric such as cost per unit that shows a counted benefit is happening; it is never added to it. Total Business Impact
Switching cost: What it costs to leave a vendor or a system: migration, rebuilt history, parallel running and lost terms. Build vs Buy Economics and Investment Decisions
Switching value: The value an assumption would have to reach for the option to stop being worth doing. Building the AI Business Case
System of record: Where the business keeps the official version of something, such as the CRM, billing, ledger or ticketing system. From Copilots to AI Agents · also in Autonomous Workflows and AI-Native Organizations
System owner: One named person in a business role who answers for an AI system's outcome and can approve, change, pause or retire it. AI Roles, Ownership and Accountability

T

Tacit knowledge: Know-how people use but cannot easily put into words, such as an expert's judgment of when something looks wrong. AI in Knowledge Management
Target maturity: The level a use or unit actually needs, set by strategy, risk, scale and law, not by the top of the ladder. The AI Maturity Model
Task decomposition: Breaking a role into the tasks it actually contains, so AI's impact is judged task by task rather than by job title. AI and the Future of Work
Task-level suitability: Deciding for each task in a workflow whether a fixed rule, AI, a person or removal is the right answer. AI Use-Case Discovery and Design
Task productivity: How much faster or better one task is done with AI. An input to business results, not proof of them. AI and Workforce Productivity
Temperature: A generation setting. Lower makes output more conservative and repeatable; higher makes it more varied. Neither makes it correct. How Generative AI Works
Test-time compute: Extra computation a model spends while answering, such as reasoning step by step; it raises cost and latency per answer. Training vs Inference
Thin wrapper: A product that is mostly a rented model behind an interface; easy to copy and exposed to the provider. AI-Native Products and Business Models
Third-party AI risk: Risk from the provider of an AI service - its security, contract, support, subprocessors, viability and continuity - rather than from the model's behavior. Model and Third-Party Risk
Three clocks: Technology (a capability appears), employee (people use it) and enterprise (the organization sees, bounds and scales it). Leaders close the last gap. The Speed of AI Adoption
Three task states: Human-led (the person does it), AI-assisted (the person uses AI and stays accountable), AI-automated (AI does it; the person handles exceptions). AI and the Future of Work
Threshold: The quality, cost, speed and risk line a capability must cross for one named workflow before you act. Where AI Is Going Next
Time-boxed escalation: A rule that sends an unresolved trade-off up one level automatically, with options, once a fixed number of working days has passed. AI Transformation Governance and Executive Sponsorship
Time horizon: METR's measure: the length of task, in human expert time, that a model completes at a given success rate. Where AI Is Going Next
Token: A piece of text - a word, part of a word or punctuation - that a model reads and writes; usage is billed in tokens. Tokens, Context and Embeddings
Total business impact: The combined effect of an initiative across all dimensions, after dependencies, the counterfactual and double counting are addressed. Total Business Impact
Total cost of ownership: Everything it costs to build, run, operate and change an AI system over its whole life, including migration and retirement. Understanding AI Total Cost of Ownership
Trace-back test: Checking that an AI initiative links up through a priority and the AI strategy to the business strategy before it is funded as a bet. What Is an Enterprise AI Strategy?
Tracking tag: Code on a website that sends visitor actions to an ad platform, whose AI uses them to target and optimize ads. AI in Sales and Marketing
Trade secret: Valuable information that is not generally known and is protected only while its owner takes reasonable measures to keep it secret. Intellectual Property and Copyright
Traditional AI: Task-specific systems that predict, classify or optimize, usually built into a business application rather than used directly. Generative AI Changes the Game
Training: Repeatedly adjusting a model's parameters on example data until its outputs improve; this creates the model's capability. Training vs Inference
Training at scale: Teaching a model on vast data with large clusters of specialized chips; at the frontier, its cost keeps rising. Why AI, Why Now?
Training data: Historical examples used to build or adapt a model; they decide which patterns it can learn. Data, Models and Compute
Transformation: Redesigning the process, roles, controls and measures around what AI makes possible, not just adding a tool. AI Is Changing Everything
Transformation board: A small executive forum that decides the AI portfolio. Its authority test: it can move money, stop initiatives and re-sequence the roadmap. AI Transformation Governance and Executive Sponsorship
Transformation governance: How leadership sets direction, allocates money and capacity, settles trade-offs between functions, accepts material risk and removes blockers so the roadmap moves. AI Transformation Governance and Executive Sponsorship
Transformation initiative: Multi-year change to how part of the enterprise creates value: its processes, roles, decision rights, data and systems. Quick Wins, Strategic Bets and Transformation Initiatives
Trigger: A threshold on a monthly number, written into the approval, that reopens the business case when crossed. Module 08 Synthesis — Making AI Economically Sustainable

U

Umbrella versus rain dance: An umbrella decision needs only a prediction; a rain-dance decision tries to change the outcome and needs evidence of cause. AI in Prediction, Forecasting and Optimization
Unit cost: Total process cost divided by cases or transactions in the same period, including what AI itself costs to run. Where AI Reduces Cost
Unit of economics: The business unit a case is measured in - per case, document, customer or shipment - for both cost and value. The Economics of AI
Unsupervised learning: Finding structure, such as clusters or anomalies, in data that has no answers attached; people judge what it means. AI vs Machine Learning
Use-by-market map: A table of each AI use against each market served, showing the duties and dates that apply in each. AI, Regulation, Geopolitics and Global Competition
Use case: A capability applied to a specific task, for specific people, at a specific point in a workflow, with a decision that changes. From AI Capability to Business Outcome · also in The Enterprise AI Use-Case Landscape
Use-case chain: Role, current problem, AI capability, new workflow, measurable outcome. A proposal missing a link is not yet a use case. The Enterprise AI Use-Case Landscape
Use-case design brief: One page naming the problem and measure, four roles, workflow map, task split, data and systems, and the smallest test with a stop rule. AI Use-Case Discovery and Design
Utilization: The share of available capacity actually used for productive work. High productivity with no demand for the time leaves capacity idle. Productivity vs Realized Capacity · also in Model, Compute and Infrastructure Costs

V

Value discipline: Treacy and Wiersema's three routes to market leadership: operational excellence, customer intimacy or product leadership. Start With Business Strategy
Value hypothesis: A testable claim: if this capability, for this workflow, then this outcome improves by this much, while a constraint holds. What Does AI Value Actually Mean?
Value of information: What reducing an uncertainty is worth to a decision; for a go-or-stop choice, at most the chance of failure times the loss avoided. Experimentation vs Production Economics
Value pool: A place where the business creates, captures or loses value - revenue, cost, customer or risk. Finding Strategic AI Opportunities
Vanity metric: A count, such as licenses or prompts sent, that rises by itself and guides no decision. Leading vs Lagging AI Metrics

W

Weak-link hypothesis: A production chain moves at the speed of its slowest step, so faster code writing is absorbed by human review and release. AI in Software Engineering
Weakest link: The dimension whose gap limits the whole ambition, however strong the others are; after Kremer's O-ring theory. Assessing Enterprise AI Readiness
Web-scale data: The public internet's text, code and images used as training material; it does not include your company's private information. Why AI, Why Now?
Within-user learning: A product learns one customer's habits. It helps that customer, plateaus quickly and at best raises switching costs. Proprietary Data, AI Moats and Differentiation
Workflow integration: AI wired into the systems, hand-offs and sign-offs where work happens, rather than used in a separate chat tab. The AI Competitive Advantage · also in AI Strategy vs AI Adoption
Works council: An elected employee body which, in several European countries, must be informed, consulted or asked to agree before monitoring technology is introduced. AI Change Management, KPIs and Operating Rhythm