Glossary
Decagon’s AI and CX glossary
The world of agentic AI and modern customer experience moves fast. This glossary breaks down the most important concepts in plain language, so your team can evaluate, deploy, and scale CX platforms with confidence.
Guardrail evaluation
H
Hallucination detection
Help desk
Hill climbing
Human-in-the-loop (HITL)
I
Inference time
Intelligent Virtual Agent (IVAs)
Intent detection
Intent recognition
Interactive voice response (IVR)
ISO 27001
ISO 42001
J
K
Knowledge base
Knowledge graph
KYC in customer service
L
LangGraph
Large language model
Latency
Lifetime value (LTV)
Live chat
LLM router
LLM token usage
M
Mean opinion score (MOS)
Mixture of experts
Model card
Model context protocol
Model drift
Multi-channel vs omnichannel
Multimodal AI
Multi-turn conversation
N
Named entity recognition
Natural language processing (NLP)
Net Promoter Score (NPS)
Next-best action
NLG (natural language generation)
NLU (natural language understanding)
O
Omnichannel customer support
Outbound voice AI
P
PII redaction
Predictive dialer
Proactive customer support
Prompt engineering
Prompt injection
Prompt versioning
Prosody
Public Switched Telephone Network (PSTN)
Q
QA in customer service
Queue management
R
Reinforcement learning
Resolution-based pricing
Resolution rate
Responsible AI
Retrieval augmented generation (RAG)
S
Self-service rate
Semantic search
Sentiment analysis
Service level agreement (SLA)
Session border controller
Short message service (SMS)
SIP transfer
Skill-based routing
SLA compliance rate
SOC 2 Type II
Speaker diarization
Speech synthesis
Speech to intent
Speech to speech
Structured output
T
Telephone Consumer Protection Act (TCPA)
Telephony
Text classification
Ticket deflection
Ticketing system
Ticket prioritization
Ticket routing
Ticket volume
Tiered support
Token limit
Tone of voice in AI
Topic modeling
Transfer learning
Turn-taking (conversational AI)
U
Utterance
V
Vector embedding
Voice activity detection (VAD)
