Glossary
Definitions and concepts for AI in customer service: AI agents, deflection, AHT, FRT, knowledge bases, RAG, hallucination and more.
71 terms
AI fundamentals
Agentic AI
Agentic AI describes systems that pursue multi-step goals autonomously, taking actions and adapting to results rather than responding turn-by-turn.
Learn moreAI agent
An AI agent is a software system that perceives a goal, plans steps toward it, takes actions in tools or APIs, and adapts based on results.
Learn moreAI guardrails
AI guardrails are policy and verification layers that constrain what an AI system will say or do, reducing hallucinations and unsafe outputs.
Learn moreAI hallucination
An AI hallucination is when a model produces a confident, fluent, but factually incorrect or fabricated response.
Learn moreContext window
The maximum amount of text, measured in tokens, that an AI model can read and consider at one time when generating a response.
Learn moreConversational AI
Conversational AI is the category of software that uses natural language to interact with users, voice, chat, or messaging, going beyond scripted responses.
Learn moreEmbeddings
Embeddings are numeric representations of text or other data that place items with similar meaning close together so machines can compare them.
Learn moreFine-tuning
Fine-tuning is the process of further training a pretrained model on a focused dataset so it performs better on a specific task or domain.
Learn moreFunction calling
Function calling lets an AI model trigger external tools or actions, like looking up an order, by returning a structured request the system can run.
Learn moreGenerative AI
Generative AI is software that produces new content, text, images, code, audio, rather than classifying or scoring existing data.
Learn moreHuman in the loop
Human in the loop is a design where people review, approve, or correct an AI system's outputs so humans stay in control of important decisions.
Learn moreLLM (Large Language Model)
A large language model is a neural network trained on massive amounts of text that predicts the next token given context.
Learn moreMachine learning
Machine learning is a branch of AI where systems learn patterns from data and improve at a task without being explicitly programmed for every case.
Learn moreMulti-agent system
A multi-agent system orchestrates several specialized AI agents that collaborate to solve problems too complex for a single agent.
Learn moreNatural language processing
Natural language processing is the field of AI that lets computers read, understand, and produce human language in text or speech.
Learn moreNeural network
A neural network is a machine learning model made of layered, connected units that learn to map inputs to outputs by adjusting weighted connections.
Learn moreRAG (Retrieval-Augmented Generation)
RAG is a pattern where an AI model retrieves relevant documents from a knowledge base and uses them as context to generate a response.
Learn moreSemantic search
Semantic search finds results based on meaning rather than exact keywords, so a query returns relevant content even when the wording differs.
Learn moreTokens
Tokens are the small chunks of text, often words or parts of words, that an AI model reads and generates when processing language.
Learn moreTransformer model
A transformer model is a neural network design that uses attention to weigh how words in a sequence relate, and it underpins most modern language models.
Learn moreVector database
A vector database stores embeddings and finds the items most similar in meaning to a query, making search by meaning fast at large scale.
Learn moreAI for support
AI agent assist (Copilot)
AI agent assist (or Copilot) is a real-time helper that suggests replies, surfaces knowledge, and recommends next-best actions while a human agent works on a ticket.
Learn moreAI chatbot
An AI chatbot is a conversational interface, usually web or mobile chat, that uses AI to interact with users in natural language.
Learn moreAI copilot
An AI copilot is an assistant that works alongside a support agent, suggesting replies, surfacing answers, and handling routine steps in real time.
Learn moreAI customer service
AI customer service is the use of AI agents, copilots, and automation to resolve customer issues across channels.
Learn moreAutomated resolution
Automated resolution is when a support request is fully solved by software, without a human agent stepping in.
Learn moreAutomated ticket triage
Automated ticket triage uses AI to read incoming support requests and sort, prioritize, and route them to the right team or agent.
Learn moreCustomer support automation
Customer support automation is the use of software (AI agents, workflows, rules) to handle support tasks without human intervention.
Learn moreIntent recognition
Intent recognition is the process of classifying what a user wants from a natural-language input, the foundation of routing and resolution.
Learn moreMultilingual support
Multilingual support is the ability to help customers in their preferred language across chat, email, and voice channels.
Learn moreNatural Language Understanding (NLU)
NLU is the field of AI focused on parsing meaning from natural-language input, intents, entities, sentiment, and context.
Learn moreOmnichannel support
Omnichannel support delivers a unified customer experience across all channels, email, chat, voice, social, in-app, with shared context.
Learn moreProactive support
Proactive support reaches out to customers before they ask for help, addressing likely issues based on signals and known patterns.
Learn morePrompt engineering
Prompt engineering is the practice of designing inputs to LLMs to elicit reliable, useful outputs in production.
Learn moreSelf-service support
Self-service support is when customers resolve their own issues through documentation, chatbots, or community, without contacting an agent.
Learn moreSentiment analysis
Sentiment analysis uses AI to detect the emotional tone of a message, such as positive, negative, or neutral, from the customer's words.
Learn moreTicket summarization
Ticket summarization uses AI to condense a support conversation into a short, clear recap of the issue, context, and resolution.
Learn moreVoice AI
Voice AI is the application of AI to spoken-language interfaces, voice agents, contact-center automation, and IVR replacement.
Learn moreWorkflow automation
Workflow automation uses software to run multi-step support processes automatically, moving tickets and data between systems without manual handoffs.
Learn moreSupport metrics
Abandonment rate
Abandonment rate is the share of customers who leave a queue before reaching an agent, typically while waiting on call or chat.
Learn moreAgent occupancy
Agent occupancy is the share of an agent's logged-in time spent actively handling contacts rather than waiting for work.
Learn moreAverage Handle Time (AHT)
AHT is the average minutes a support agent spends handling a ticket from open to close, the operational cost-per-contact metric.
Learn moreAverage resolution time
Average resolution time is the mean time it takes to fully resolve a ticket, measured from creation to closure.
Learn moreContainment rate
Containment rate is the share of contacts fully handled by automation without ever being escalated to a human agent.
Learn moreCost per contact
Cost per contact is the total support spend divided by the number of customer contacts handled in a given period.
Learn moreCSAT (Customer Satisfaction Score)
CSAT is the score customers give after an interaction, usually a 1-5 scale rating overall satisfaction with the support experience.
Learn moreCustomer Effort Score
Customer Effort Score measures how easy it was for a customer to get their issue resolved, based on an ease-of-resolution rating.
Learn moreDeflection rate
Deflection rate is the percentage of customer issues resolved without human agent involvement. The most-quoted metric in AI-for-support sales.
Learn moreFirst Contact Resolution (FCR)
FCR is the share of tickets resolved in the customer's first interaction, without follow-ups, re-opens, or repeat contacts.
Learn moreFirst Response Time (FRT)
FRT is the time from when a customer submits a ticket to when they receive the first meaningful response. The customer-facing speed metric.
Learn moreNet Promoter Score
Net Promoter Score measures customer loyalty by asking how likely customers are to recommend a company, then subtracting detractors from promoters.
Learn moreSLA compliance
SLA compliance is the share of tickets handled within the response and resolution targets set in a service level agreement.
Learn moreTicket backlog
Ticket backlog is the count of open, unresolved tickets that have accumulated and still await action at a point in time.
Learn moreTicket deflection
Ticket deflection is the share of customer issues resolved without human agent involvement, through AI, self-service, or automation.
Learn moreVoice of the customer
The practice of gathering and acting on customer feedback across channels to understand their needs, expectations, and experience.
Learn moreSupport operations
Canned responses
Canned responses are pre-written reply templates agents reuse to answer common questions quickly and consistently.
Learn moreEscalation management
Escalation management is the process of moving a support case to a higher tier or specialist when it cannot be resolved at the current level.
Learn moreHelpdesk software
Helpdesk software is the platform support teams use to receive, track, and resolve customer tickets across channels.
Learn moreITSM (IT Service Management)
ITSM is the discipline and software category for running IT operations as a service to the business, incidents, changes, problems, assets.
Learn moreKnowledge base
A knowledge base is the repository of articles, FAQs, and documentation that supports self-service and AI-driven resolution.
Learn moreKnowledge management system
A knowledge management system is the platform plus operating loop that keeps a knowledge base current, accurate, and useful.
Learn moreKnowledge-Centered Service
A support methodology where teams capture and improve knowledge as a natural part of solving tickets, so answers are reused instead of rebuilt each time.
Learn moreQueue management
The practice of organizing and prioritizing incoming support tickets so the right work reaches the right agent in the right order.
Learn moreService desk
A service desk is the internal IT support function, and the software platform that runs it, handling employee tickets and IT operations.
Learn moreService Level Agreement
A documented commitment between a support provider and its customers that defines target response and resolution times and the consequences of missing them.
Learn moreShift-left support
A strategy that moves issue resolution closer to the customer through self service and frontline tools, reducing the need to escalate to specialists.
Learn moreSupport quality assurance
The process of reviewing support interactions against a defined standard to measure quality, give feedback, and improve how agents serve customers.
Learn moreTicket routing
Ticket routing is the process of directing each incoming support ticket to the right agent, team, or queue for handling.
Learn moreTiered support
A support model that organizes agents into levels, where simpler issues are handled first and harder ones escalate to more specialized teams.
Learn moreWorkforce management
The discipline of forecasting support demand and scheduling the right number of agents so service targets are met without overstaffing.
Learn more