AA Essentials key terms

Using technology to carry out a task or process with little or no human intervention, following rules that have been defined in advance so results are repeatable.

An ordered series of steps that moves work from start to finish, including the decisions, handoffs, and approvals needed to complete a business outcome.

A software program configured to perform defined actions on behalf of a user or system, such as clicking, typing, reading data, or calling another service.

A single unit of work with a clear start and end, such as opening a file or sending an email, that can be combined with other tasks into a larger workflow.

Decision making that follows explicit if-then conditions written by a person, so the same input always produces the same branch and outcome.

A process whose steps and outcomes are predictable for a given input, making it a strong candidate for straightforward automation without judgment.

A mode where a bot runs on a user's machine and is triggered by that person, typically handling small steps while the human stays in control.

A mode where bots run on their own, often on dedicated servers and on a schedule, without a person present to start or supervise each run.

A design where a person reviews, approves, or corrects a step in an automated flow before the process continues, usually for exceptions or high-risk actions.

The planned response when a process hits an unexpected condition, such as missing data or a failed system, including logging, retrying, or escalating to a person.

The practice of choosing which processes to automate first, weighing factors such as volume, rule stability, error rates, and effort saved.

Analyzing system event logs to reconstruct how a process actually runs, revealing bottlenecks, rework, and variation that may differ from the documented procedure.

Coordinating multiple bots, scripts, and systems so their steps run in the right order, with shared status and handoffs managed centrally.

A central control component that schedules, starts, monitors, and tracks bots and their jobs across an environment.

A holding list of work items that bots pull from in order, allowing many items to be processed reliably and progress to be tracked.

The field of building computer systems that perform tasks associated with human intelligence, such as recognizing patterns, understanding language, or making predictions.

A branch of AI where a model learns patterns from example data rather than being given explicit rules, then uses those patterns to make predictions on new data.

The trained artifact produced by a machine learning process that takes an input and produces a prediction, classification, or generated output.

The labeled or unlabeled examples used to teach a model, whose quality, coverage, and representativeness strongly shape how well the model performs.

A training approach that uses examples paired with correct answers, so the model learns to map inputs to known labels such as spam or not spam.

A prediction task that assigns an input to one of a fixed set of categories, such as routing an email to a department or flagging a document as urgent.

A model trained on very large amounts of text that can generate and interpret language, supporting tasks such as summarizing, drafting, and answering questions.

AI systems that create new content, such as text, images, or code, in response to an instruction, rather than only labeling or scoring existing data.

The instruction or input text given to a language model that shapes its response, including any context, examples, or formatting guidance supplied.

The practice of designing and refining prompts, including context and examples, to get more accurate, consistent, and useful outputs from a language model.

A confident but incorrect or fabricated output from an AI model, which is why generated content needs verification before it drives real decisions.

The act of running a trained model on new input to produce an output, as opposed to the earlier training phase that built the model.

A number a model returns indicating how sure it is about a prediction, often used to route low-certainty cases to a person for review.

A systematic skew in model outputs caused by unbalanced or unrepresentative training data or design choices, which can produce unfair or inaccurate results.

A set of practices for building and using AI that considers fairness, transparency, privacy, safety, and accountability to the people affected.

A connection that lets two systems exchange data or trigger actions, often through an API, a file exchange, or a user interface.

A defined interface that lets one piece of software request data or actions from another in a structured way, without needing to operate its user interface.

A specific address within an API that accepts requests for a particular resource or action, such as retrieving a customer record or creating a ticket.

A mechanism where one system automatically sends data to another when an event occurs, letting automation start immediately instead of polling on a schedule.

A prebuilt component that handles the details of talking to a particular system, so a builder can read or write its data without custom code.

The event that starts an automated run, such as a new email, a form submission, a schedule time, or a change in a database record.

A time-based setting that starts an automation at set intervals or times, used when the work is periodic rather than event-driven.

A named placeholder that stores a value during a run so later steps can read or update it, such as a customer name or an order total.

A test inside an automation that checks whether something is true, then chooses which path to follow based on the result.

A structure that repeats a set of steps for each item in a collection or until a stopping condition is met.

Defining how a field in one system corresponds to a field in another, so values are placed in the right place during an exchange.

An attribute or path used to identify a specific element on a screen or page so a bot can reliably find and act on it.

Driving an application through its user interface by simulating clicks and typing, used when no API is available for integration.

A record of failures and warnings produced during automated runs, used to diagnose problems and improve reliability.

A point in a workflow where a designated person must confirm or reject before the process moves forward, often used for spending or sensitive changes.

The process of proving an identity to a system, such as with a password, token, or certificate, so only permitted parties can access it.

A secret used to prove identity, such as a username and password, key, or token, which automations must store and handle securely.

The value gained from an automation compared with its cost to build and maintain, often measured in hours saved, errors reduced, or faster cycle time.

The full cost of an automation over its life, including design, build, licensing, infrastructure, monitoring, maintenance, and eventual retirement.

A central team that sets standards, provides reusable components and training, and governs how automation is designed and deployed across an organization.

The rules, roles, and review processes that decide how automations are approved, built, secured, monitored, and retired within an organization.

A business user who builds automations or apps with low-code tools, usually within guardrails set by IT and a center of excellence.

A development approach that uses visual builders and prebuilt components so people can create automations with less hand-written code.

A chronological record of what an automation did, when, and on whose behalf, supporting troubleshooting, compliance checks, and accountability.

Rules that determine which people and bots may view or change specific systems and data, limiting risk from mistakes or misuse.

Handling personal and sensitive information in line with laws and policy, including limiting what is collected, who sees it, and how long it is kept.

The ability of an automation setup to handle growing volumes of work by adding bot capacity or resources without redesigning the process.

The amount of work an automation completes in a given period, such as items processed per hour, used to judge capacity and performance.

The delay between a request or event and the response or completed action, which affects how quickly automated steps feel and perform.

The structured work of preparing people for a new automated process, including communication, training, and support so adoption succeeds.