AI chatbot glossary. 46 terms, in plain words.
Short, neutral definitions of the terms you meet when choosing or running a website chatbot: how the AI works, how support teams measure it, and how it fits on your site.
A
- AI agentAn AI agent is an AI system that works toward a goal by deciding on steps and taking actions, such as calling tools, searching data or updating records, rather than only producing a reply.
- AI chatbotAn AI chatbot is software that holds a text conversation with people and writes its replies using artificial intelligence, usually a large language model, instead of following a fixed script.
- AI disclosureAI disclosure means telling people when they are interacting with an AI system rather than a human.
- AI OverviewsAI Overviews are AI-generated summaries that Google Search shows above the regular results for some queries, with links to the web pages the summary draws on.
- Answer engine optimization (AEO)Answer engine optimization (AEO) is the practice of structuring web content so AI answer engines, such as ChatGPT, Perplexity and Google AI Overviews, can find, understand and cite it.
- Average handle time (AHT)Average handle time (AHT) is the average time a support agent spends on one customer interaction, from start to finish, including talking or chatting, any hold time and the follow-up work afterward.
C
- Chat widgetA chat widget is the small interface embedded on a website, usually a round button in a bottom corner, that opens a chat window when clicked.
- ChunkingChunking is splitting long documents into smaller passages, or chunks, before creating embeddings for retrieval.
- Containment rateContainment rate is the percentage of conversations that a chatbot or automated phone system handles from start to finish without transferring the customer to a human.
- Context windowA context window is the maximum amount of text, measured in tokens, that a language model can take into account at once.
- Conversation inboxA conversation inbox is the place in a support or chat tool where a team can see every customer conversation in one list, usually with the full transcript, time, status and any contact details.
- Conversational AIConversational AI is the field of technology that lets computers hold natural conversations with people in text or speech.
- Core Web VitalsCore Web Vitals are three metrics Google uses to measure real-world page experience: Largest Contentful Paint (loading speed), Interaction to Next Paint (responsiveness) and Cumulative Layout Shift (visual stability).
- CSAT (customer satisfaction score)CSAT, or customer satisfaction score, measures how satisfied customers are with a specific interaction or service, usually through a one-question survey after a conversation.
F
- Fallback messageA fallback message is what a chatbot says when it can't understand a request or doesn't have an answer.
- FAQ pageAn FAQ page lists frequently asked questions about a business, product or topic, each followed by a short answer.
- Fine-tuningFine-tuning is further training of a pre-trained AI model on a smaller, focused dataset to adjust its behavior, style or skill at a specific task.
- First response time (FRT)First response time (FRT) is the time between a customer contacting support and receiving the first reply from a person or system.
G
- Greeting messageA greeting message is the first message a chatbot shows when a visitor opens the chat window, before they type anything.
- GroundingGrounding means tying an AI model's answer to a specific, trusted source of information, such as retrieved documents, a database or search results, so its claims can be traced and checked.
- GuardrailsGuardrails are rules and checks placed around an AI system to keep its behavior within safe, intended limits.
H
- HallucinationA hallucination is an output from an AI model that sounds confident and plausible but is false or unsupported by its sources, such as an invented fact, quote, price, policy or link.
- Help centerA help center is a section of a website, or a separate site, that collects support articles, guides and answers to common questions, usually organized by topic and searchable.
- Human handoffHuman handoff is the point where an automated chatbot passes a conversation to a person, either because the customer asks for one or because the bot can't help.
L
- Large language model (LLM)A large language model (LLM) is an AI model trained on very large amounts of text to predict and generate language.
- Lead captureLead capture is the process of collecting contact details, such as a name, email address or phone number, from potential customers so a business can follow up.
- Live chatLive chat is a website or in-app messaging channel where customers chat in real time with human support or sales agents.
- llms.txtllms.txt is a proposed web standard for a Markdown file placed at a site's root, at /llms.txt, that gives large language models a concise summary of the site and links to its most useful pages.
P
- Per-resolution pricingPer-resolution pricing is a billing model in which an AI support vendor charges a fee for each conversation its AI resolves, rather than a flat monthly price or a charge per message.
- Prompt injectionPrompt injection is an attack in which someone hides instructions in text that an AI model reads, such as a chat message, web page or document, to make it ignore its original instructions.
R
- Resolution rateResolution rate is the percentage of conversations or tickets in which the customer's issue was actually solved.
- Retrieval-augmented generation (RAG)Retrieval-augmented generation (RAG) is a technique in which an AI system first retrieves relevant passages from a chosen set of documents, then gives them to a language model to write its answer.
S
- Script tagA script tag is the HTML <script> element, which loads or runs JavaScript on a web page.
- Self-serviceSelf-service in customer support means giving customers the tools to find answers and solve problems on their own, without contacting staff.
- Shadow DOMShadow DOM is a web standard that lets a component attach its own hidden, encapsulated tree of HTML elements to a page.
- System promptA system prompt is a set of instructions given to a language model before the conversation starts, which sets its role, tone, rules and limits.
T
- Ticket deflectionTicket deflection is the practice of answering customer questions through self-service, such as a help center, FAQ or chatbot, so they never become support tickets.
- TokenA token is the unit of text a language model reads and writes: a whole word, part of a word, a number or a punctuation mark.
See how these ideas apply in practice in the use-case guides, or read the ChatBento blog.
An AI chatbot that answers from your site. Free to start. Paid plans from $8 a month.
14-day free trial. No card required.