AI readiness glossary: plain-English definitions

Conversations about AI are full of terms that vendors, consultants and managers use in different ways. This glossary gives a short, plain-English definition of the terms SME leaders meet most often. It covers AI itself, readiness, processes, systems and data.

Get a free Snapshot→Go to the terms

How to use it

Short definitions, grouped by topic.

Each definition is written to be read on its own, in one or two sentences. Where a term has its own guide, the definition links to it. The terms are grouped in the order a business usually meets them: first AI, then readiness, then the three foundations of processes, systems and data.

AI basics

The AI terms you will hear most.

Artificial intelligence (AI): Computer systems that perform tasks which normally need human judgement, such as recognising patterns, answering questions or making predictions. In business, AI is usually built into software rather than bought as a separate product.
Machine learning: A type of AI that learns patterns from past data instead of following fixed rules. A demand forecast that improves as more sales history is added is an example.
Generative AI: AI that produces new content, such as text, images or summaries, in response to a request. Chat assistants are the most common example.
Large language model (LLM): The type of AI model behind most chat assistants. It is trained on very large amounts of text, so it can read and write in everyday language.
AI agent: An AI system that can carry out a series of steps on its own to reach a goal, such as checking stock and then drafting a purchase order. Because it acts as well as answers, it needs clearer limits and approvals than a chat assistant. Also called agentic AI.
Hallucination: An answer from an AI tool that sounds confident but is wrong or invented. It is the main reason AI outputs need checking, especially when they are used for decisions.
AI use case: One specific business task that AI is applied to, such as matching invoices to payments. Readiness is easier to judge one use case at a time than for the whole business at once.
AI pilot: A small, time-limited trial of one AI use case in one team or process, with a clear before-and-after measure. A pilot shows whether the use case works before the business spends more on it.

Readiness

Terms used to describe readiness.

AI readiness: How prepared a business is to adopt AI and get lasting value from it. It covers strategy, leadership, skills, processes, systems, data and governance. Our guide explains what AI readiness is in more detail.
AI readiness assessment: A structured review of how prepared a business is for AI, based on evidence such as interviews, documents and system data. It should end with a prioritised list of what to fix first. See what an assessment involves.
Maturity level: A step on a scale that describes how developed a capability is, from informal to routinely improved. The SEL Foundation Index uses four levels: Ad-hoc, Defined, Integrated and Optimising.
SEL Foundation Index: SEL's diagnostic framework. It scores a business's process, system and data foundations across five dimensions and four maturity levels, and shows which gap to close first. See how the Index works.
Senior sponsor: The senior person who is responsible for the result of an AI or improvement project. The sponsor decides priorities, approves spending and keeps the work going after the first enthusiasm fades.
AI governance: The rules and responsibilities that decide how a business uses AI. It covers who approves each use, what data the AI may use, how results are checked, and what happens when something goes wrong.

Processes

Process terms.

Business process: A repeated sequence of steps that produces a result, such as taking an order through to collecting payment. Processes are the first of the three foundations, because systems and data both depend on them.
Process mapping: Drawing or writing down each step of a process, who does it and where it goes wrong. It is the usual starting point before any automation. See our guide to process mapping.
Standard operating procedure (SOP): A written description of how a task should be done, so that anyone trained can do it the same way. SOPs make a process consistent enough to measure and automate.
Process owner: The person responsible for how a process performs and for improving it. Without an owner, problems are noticed but not fixed.
Automation: Using software to carry out steps that people did by hand, such as sending reminders or copying data between systems. Automation works well only on a process that already runs the same way each time.
Workaround: An unofficial way of getting work done when a process or system does not fit, such as keeping a side spreadsheet. Workarounds are a common sign that foundations need attention.

Systems

System terms.

ERP (enterprise resource planning): A system that combines finance, purchasing, inventory, sales and often production in one shared database. Many SMEs use separate connected systems instead. See how to choose business software.
CRM (customer relationship management): A system that keeps customer details, sales opportunities, quotes and follow-ups in one place.
POS (point of sale): The system that records sales at the counter or table and takes payment. It often tracks stock at each outlet as well.
System of record: The one system that is treated as the official source for a type of data. For example, the accounting system is usually the system of record for invoices.
Integration: A connection that lets two systems exchange data automatically, so that nobody has to re-enter it by hand.
API (application programming interface): A standard way for one piece of software to request data from, or send data to, another. Most modern integrations use APIs.
Cloud software (SaaS): Software that runs on the provider's servers and is used through a browser or app, usually for a monthly subscription. SaaS stands for software as a service.

Data

Data terms.

Data quality: How accurate, complete, consistent and up to date a set of data is. Our guide on preparing data for AI includes a ten-point check.
Master data: The core records that many processes share, such as customers, products, suppliers and employees. Errors in master data spread into every report and AI tool that uses it.
Data definition: An agreed meaning for a business term, such as “active customer” or “gross margin”. A list of these definitions is often called a data dictionary.
Data owner: The person responsible for the quality of one set of data, such as product records. The data owner decides how the data is defined and who may change it.
Single source of truth: One agreed place where each key figure comes from, so that every team works from the same numbers. See do SMEs need a data warehouse?
Data warehouse: A central database that collects data from several systems and organises it for reporting and analysis. It is one way to create a single source of truth, but not the only way.
Business intelligence (BI): Software that turns data into reports and dashboards. A dashboard is a single screen that shows the key figures for a team or the business.
Data governance: The rules and responsibilities for how data is defined, stored, accessed and protected. In an SME, it can be as simple as named owners, agreed definitions and access by role.
Personal data protection law: The law that controls how a business collects and uses information about people. In Singapore and Malaysia it is the Personal Data Protection Act (PDPA). In Indonesia it is Law 27 of 2022 (UU PDP).

Common questions

Questions about the glossary.

Why does a business need a shared glossary for AI?

Because the same word often means different things to different people. A vendor, a manager and a consultant may each mean something different by “integration” or “AI agent”. Agreeing the meaning first avoids buying, or approving, the wrong thing.

Is AI readiness the same as digital transformation?

No, although the two overlap. Digital transformation is about adopting systems that suit the business, which is one part of AI readiness. AI readiness also depends on defined processes, reliable data, skilled people and clear leadership.

Which of these terms matter most for a first AI project?

Use case, process owner, system of record, data definition and senior sponsor. If each of these is clear for your first project, most of the common causes of failure are already dealt with.

GuideWhat is AI readiness, and how do you know if you have it?GuideDo SMEs need a data warehouse?

Find out where you stand

See how ready your foundations are.

The free Readiness Snapshot takes about five minutes. It emails you a PDF report showing where your processes, systems and data stand on the path to AI, and which gap to close first.

Get a free Snapshot→Contact us