Plain English

AI glossary for business owners

The AI terms you keep hearing, explained the way I'd explain them across a table. No computer science degree needed. Use the search box to jump to a term.

Terms, A to Z

AI agent

An AI tool that can carry out a multi-step task on a computer, not just answer a question. It can read an email, look something up, update your CRM and draft a reply, in that order. More on AI agents.

Agentic AI

The buzzword for AI that takes actions instead of just chatting. Same idea as an AI agent.

AI automation

Any repeatable process where AI does part of the work: reading, writing, sorting or deciding. Usually runs on a schedule or a trigger.

AI manager (fractional)

A person who directs AI agents for a business part-time: assigns work, gives the agents access and instructions, and checks the output. What a fractional AI manager does.

API

The way two pieces of software talk to each other directly. When people say two apps "integrate", there's usually an API underneath.

Chatbot

A program that answers questions in a chat window. Older chatbots followed scripts. Newer ones use an LLM.

ChatGPT, Claude, Gemini

The three best-known AI assistants, made by OpenAI, Anthropic and Google. Each is good at slightly different things. Which should my business use?

Context

The information an AI has in front of it when it works: your instructions, documents, examples. Better context, better results.

Context window

How much information an AI can hold in mind at once. Bigger windows let it read longer documents in one go.

CRM

Customer relationship management software. Where your contacts, leads and deals live (HubSpot, GoHighLevel, Salesforce and others). CRM cleanup and automation.

Data enrichment

Filling in missing details on a list (emails, phone numbers, websites, company size) from public sources. A common first job for an agent cleaning up a lead list.

Fine-tuning

Retraining an AI model on your own examples. Small businesses rarely need it; good instructions and context usually do the job.

Generative AI

AI that creates new text, images, audio or code instead of only sorting or predicting. ChatGPT, Claude and Gemini are all generative AI.

Guardrails

Rules that limit what an AI agent can do or say: what it can access, what needs human approval, topics it should avoid. How I set them.

Hallucination

When an AI states something false with confidence. The reason a human should check anything that matters. What happens when AI makes a mistake?

Human in the loop

A setup where a person reviews or approves an AI's work before it counts. How I run every client account.

Integration

A connection between two apps so information moves between them automatically.

Knowledge base

A collection of your documents (SOPs, FAQs, policies) that an AI can search to answer questions accurately.

LLM (large language model)

The engine behind tools like ChatGPT and Claude. Trained on huge amounts of text to read and write language.

MCP (Model Context Protocol)

A newer standard that lets AI assistants plug into business tools (email, drive, CRM) in a consistent way.

Model

A specific version of an AI engine. Different models trade off speed, cost and smarts.

Multimodal

An AI that can work with more than text: images, audio, video, PDFs.

n8n, Zapier, Make

Workflow tools that connect apps and run automations. I use n8n. AI vs. automation vs. Zapier, explained.

No-code / low-code

Building software or automations with visual tools instead of writing code from scratch.

Prompt

The instruction you give an AI. A good prompt says who it is, what to do, what good looks like and what to avoid.

Prompt engineering

The skill of writing prompts that get reliable results. Less mystical than it sounds, but it takes practice.

Prompt injection

A trick where text hidden in an email, web page or document tries to give an AI agent new instructions. One more reason agents get limited access and a person checks anything that matters.

RAG (retrieval-augmented generation)

A way of having an AI look up your documents before it answers, so it answers from your facts instead of guessing.

SaaS

Software as a service. Any software you pay for monthly and use in a browser. The SaaS cut sheet.

SOP

Standard operating procedure. A written how-to. Agents follow good SOPs very well, which is why writing them down pays off twice. AI SOPs and training.

Token

The unit AI tools use to measure text, roughly three-quarters of a word. Usage-based AI pricing is counted in tokens.

Trigger

The event that starts an automation: a form is submitted, an email arrives, it's 8 a.m. Monday.

Voice agent

An AI agent that talks on the phone. It listens, understands and responds in a natural voice. The number on my site is one. Call it.

Webhook

A way for one app to instantly notify another when something happens, like "a new lead just came in".

Workflow

A series of steps that happen in order. An automated workflow runs those steps without someone clicking through them.

Still confused by a term?

Call my AI agent and ask it. It's patient. Or book a call and ask me.