Katexs
Glossary

Concepts

Glossary

The terms used throughout the Katexs docs and dashboard.

TermMeaning
AgentA configured AI that talks to customers. Voice agents speak; chat agents write. Also called an assistant in some API fields.
WorkspaceYour isolated tenant. Owns agents, numbers, calls, contacts, knowledge and credits.
Voice engineThe realtime layer that handles telephony, transcription, the model loop and speech synthesis.
TranscriberSpeech-to-text provider and model (STT).
VoiceText-to-speech provider and voice (TTS).
ModelThe language model that decides what the agent says and does.
First messageWhat a voice agent says when the call connects.
EndpointingHow the agent decides the caller has finished speaking.
Barge-in / interruptionThe caller speaking over the agent, which makes it stop and listen.
ToolSomething the agent can do: check a calendar, book, transfer, call your API.
Custom functionA tool you define with a JSON schema and an HTTPS endpoint.
Server URLThe HTTPS endpoint that receives tool requests for an agent's custom functions.
EventA message about a call, such as status-update or end-of-call-report.
End reasonWhy a call ended, such as customer-ended-call or silence-timed-out.
CreditsThe unit Katexs bills usage in. Voice is priced per minute, chat per message.
LedgerThe append-only record of every credit grant and charge.
Shared memoryThe cross-channel contact timeline agents can read.
CrewSeveral agents on one conversation, handing off with context.
FlowA visual, multi-step conversation graph for chat agents.
Public idThe publishable identifier of a deployed chat agent, used by the widget.
Visitor tokenA signed cookie that keeps a website visitor's chat session for 30 days.
TraceA timed record of one call, message or tool run, with its steps and cost.
BYOKBring your own key: using your own model or voice provider key in a workspace.

Last updated November 2026 · Found a problem? Email support@katexs.com