Glossary
This glossary explains the product terms that matter in day-to-day operation.
A
Add Case
The action in Conversations that turns a real situation into a reusable case in Test Suite.
Agent
The AI system you configure, test, and publish for a defined job.
Agent Version
A saved state of an Agent. Draft changes do not affect users until the intended version is published.
AI Draft
A proposed human reply generated from the available conversation context. It is never applied or sent automatically; a workspace member reviews it first.
API Key
A workspace-scoped credential used to access an Agent programmatically.
B
Brand
An identity layer used for published experiences, including visual identity and login configuration.
C
Channel
The delivery surface users interact with, such as Web, LINE, or Slack.
Conversations
The workspace view for reviewing real conversations, handling human replies, inspecting feedback, and creating cases.
Copilot
The built-in assistant that helps operators draft Agent settings, inspect failures, and suggest focused changes.
Customers
The workspace view for access mode, customer identity, tags, and per-Agent interaction mode.
E
Expert-led AI Agent
An Agent whose quality, boundaries, and release decisions remain defined by people who understand the work. AI assists with creation, testing, and improvement, but does not replace the expert's quality judgment.
H
Human Mode
The interaction mode where a person, rather than the Agent, handles replies for a specific customer-Agent pair. Manual human mode continues until someone returns it to AI.
I
Improve
The user-facing feedback action that marks a specific reply as needing work.
Instructions
The core behavior rules that define how an Agent should clarify, decide, respond, use tools, and respect boundaries.
K
Knowledge Base
Connected reference material the Agent can search when the work requires it.
L
Live Test
The place in Agent Editor for quickly trying a draft before publishing it.
M
Must-Pass Case
A case whose failure should stop a release because the behavior is central, high consequence, or part of an approved boundary.
O
Organization
The broader team boundary above individual workspaces.
R
Request Form
A structured form used when a conversation needs specific fields rather than more free-form messages.
S
Standard
The observable requirements used to judge whether a case result passed or failed.
T
Template
A structured Agent starting point. It asks setup questions and uses AI to generate a draft for review in Agent Editor.
Test Suite
The surface where teams store important cases and rerun them against Agent changes.
W
Whitelist
An access mode where only specifically allowed customers can enter the workspace experience.
Workspace
The operating boundary where Agents, Channels, Conversations, Customers, forms, and API keys are managed together.