Glossary
The words people use around AI automation, defined plainly.
Eight terms that come up in almost every conversation about automating a small business. Each definition is written to be read once and understood, not to sound impressive. If a vendor uses one of these words differently, ask them which meaning they mean.
AI automation
AI automation is software that completes a multi-step business process on its own, using an AI model to handle the judgment steps that rule-based automation cannot.
The distinction that matters is judgment. A rule-based automation can move a file when a folder changes, because that decision is a yes or no. It cannot read a rambling meeting transcript and work out which sentence was a commitment and which was thinking out loud. An AI model can, which is what makes whole processes automatable rather than just the mechanical parts of them.
In practice a working AI automation is mostly ordinary plumbing with a model doing the two or three steps in the middle that need reading comprehension. The plumbing is what makes it reliable. The model is what makes it possible.
AI voice agent
An AI voice agent is software that answers a phone call, holds a spoken conversation with the caller, and completes a task such as booking an appointment or taking intake details, handing off to a person when the call needs one.
A voice agent is three systems working together: speech recognition to turn what the caller says into text, a language model to decide what to say and do next, and speech synthesis to say it out loud. The parts that determine whether it is any good are the ones nobody demos, namely what it does when it does not understand, and how quickly it gives up and fetches a human.
The honest use case is the call you are currently missing. A voice agent that handles booking and intake reliably is worth more than one that attempts every conversation and mishandles the hard ones.
Workflow automation
Workflow automation is the practice of connecting the steps of a repeated business process so that finishing one step starts the next, without a person carrying the information across by hand.
Most small business work is already a workflow. It is just held together by somebody remembering. The value of writing it down and connecting it is not only the time saved, it is that the process stops depending on one person being at their desk.
The first useful question is never which tool to use. It is which step currently gets skipped when everyone is busy, because that is the step where automation pays for itself.
Human in the loop
Human in the loop is a design rule where an automation prepares work but a person reviews and approves it before anything leaves the business.
The rule exists because the cost of errors is not symmetrical. An automation that files a summary in the wrong folder costs a minute to fix. An automation that emails a client the wrong number costs a relationship. So the drafting is automated and the sending is not.
It also makes automation adoptable. A team that knows nothing goes out without a person seeing it will let an automation near real work much sooner than one that has to trust it completely on day one.
CRM automation
CRM automation is the practice of having contact records, deal stages, notes, and follow-up tasks update themselves from work that already happened, instead of being typed in afterwards.
The reason CRM data is usually bad is not laziness. It is that updating it is a second, unpaid job performed after the real work, by someone who already knows what happened and gains nothing personally from writing it down.
Automating the update removes the incentive problem entirely. The record becomes a byproduct of doing the work rather than a chore that follows it.
Trigger
A trigger is the event that starts an automation, such as a meeting ending, a form being submitted, or a phone call completing.
Triggers are worth naming precisely because a vague one is the most common reason an automation misfires. Starting a workflow when a calendar event ends is not the same as starting it when a recording finishes uploading, and the difference shows up as an automation that runs against an empty transcript.
A good trigger is observable, happens exactly once, and happens after everything the automation needs already exists.
Integration
An integration is a connection between two software systems that lets them exchange data automatically, so information entered in one appears in the other without being retyped.
Integrations are where automation projects usually get stuck, and rarely for interesting reasons. The common blockers are a system with no usable interface, a permission nobody can grant, or two systems that disagree about what counts as the same customer.
This is why identifying the tools a process touches is part of scoping the work rather than a detail to settle later.
AI agent
An AI agent is a program that is given a goal and a set of tools, then decides which tools to use and in what order to reach that goal, rather than following a fixed script.
The difference from a normal automation is who decides the order. A traditional automation runs the steps its author wrote, every time. An agent is told what a finished result looks like and works out the route, which makes it better at messy work and harder to predict.
For most small business processes a fixed sequence is the better answer, because the process genuinely is the same every time and predictability is worth more than flexibility. Agents earn their place where the path really does vary.
Questions
Common questions
What is the difference between AI automation and regular automation?
Regular automation follows rules its author wrote in advance, so it handles steps where the decision is a clear yes or no. AI automation adds a model that can read and interpret unstructured material like a transcript or an email, which lets it handle the judgment steps in the middle of a process that rules alone cannot.
Do I need to understand these terms to work with an automation consultant?
No. A consultant who needs you to speak their vocabulary is making their job easier rather than yours. This page exists so the words are available if you want them, not because you have to learn them first.
Which of these matters most for a small business?
Human in the loop, because it is the one that determines whether a team will actually adopt an automation. The rest describe how things are built. That one describes whether people trust the result.
Keep reading
Where these terms show up in practice
The guides
Walkthroughs of the decisions these words describe, including what to automate first and how meeting notes actually get automated.
Read the guidesThe services
The three things DeGonda Automates builds, what each one touches, and where each stops being the right answer.
See the services