Glossary

Terms are written as the app shows them. Each one links to the page that explains it.

Agents

Term	Meaning
Agent	A voice assistant that answers or makes phone calls for you. It has instructions, a voice, and optionally tools and knowledge. See What is 360° Voice Agent?
Single prompt	An agent that works from one set of instructions. See What is 360° Voice Agent?
Pathway	An agent that follows a map of steps, with a branch for each answer. Use it when the call must go in a set order. See Design a multi-step call with a pathway.
Template	A ready-made agent to start from, such as Appointment booking. See Quick start.
Persona	One line on who the agent is. It shapes how the agent speaks. See Agent settings.
Instructions	The agent’s job, its rules and how to speak, under Prompt. See Write and edit the agent’s instructions.
Greeting, Closing line	The first thing the agent says, and what it says before it ends the call. See Write and edit the agent’s instructions.
Variable	A value that changes from call to call, written as {{name}}, such as the caller’s name. It can come from an outbound call request, a tool, the phone system or test inputs. See Use variables.
Data to collect	The details the agent asks every caller for, such as a name and a date, before it reads them back. See Collect details from callers.
Tool	One of your APIs that the agent can call before, during or after a call. See Give the agent tools and Receive tool and event webhooks.
Tool credential	A key or token that a tool sends, stored encrypted. Nobody can see it again after you save it. Listed on Integrations → Tool credentials. See Give the agent tools.
Knowledge base	Documents and web pages the agent can answer questions from. See Answer questions from your documents.
Guardrail	A check on every caller turn for moments the agent must handle in a set way, for example a caller who asks for a person. Its action can transfer, end the call, warn, or just record it. A custom check is one you write. See Add guardrails.
Preset	A ready-made set of models: Budget, Balanced, Performance or Ultra. See Choose the voice and language.
Listen · think · speak, All in one	How the agent hears and speaks: three models (speech-to-text, an AI model and a voice) or one speech-to-speech model. See Choose the voice and language.
Words to listen for	Names and terms that speech-to-text should hear right. See Words to listen for.
Voice tuning	Everything that changes how a call feels, such as interruptions and turn-taking, Thinking sounds and the Maximum call length. See Agent settings.
Hold line	A short line, such as “One moment, please.”, that the agent says when a tool takes more than about a second. See What the agent says while it waits.

Drafts and versions

Term	Meaning
Draft	The agent you’re editing. Tests use it. Real calls never do. See Drafts and versions.
Version	A frozen copy of the draft, made when you publish, numbered v1, v2 and so on. See Drafts and versions.
Publish	Make a new version from the draft. It goes on the test line. See Publish and roll back versions.
Release note	What changed in a version and why, added when you publish. See Publish a new version.
Test line	The version you call to try the agent on a real phone. The newest version you published. The API calls it staging. See Test line and live line.
Live line	The version real callers reach. The API calls it production. See Test line and live line.
Put live	Move the version on the test line to the live line, with Put v3 live (or your version’s number). See Put a version live.

Calls and testing

Term	Meaning
Turn	One stretch of speech by the caller or by the agent. A transcript is a list of turns. See Read test results.
Interruption	The caller talks over the agent. In the transcript, the agent’s reply is marked “— cut off”. See Interruptions and silences.
Reply time	How long the agent takes to start answering after the caller stops talking. See Reply time.
Saved record	What the agent saved on a call once the caller confirmed it, such as a booking, lead or form. See Collect details from callers.
Transcript	What the caller and the agent said, turn by turn. See Review calls.
Outcome	How a call ended: one item from your organisation’s outcome list, chosen by the post-call analysis. See Choose outcomes.
Post-call analysis	The summary, outcome, sentiment and fields worked out after each call. See Set up post-call analysis.
p95	The value that 1 call in 20 is slower than, such as the p95 response time. See Get alerts.
Incident	What an alert rule opens when its number crosses your line. See Get alerts.
Containment	The share of calls the agent ended with a successful outcome, with no person needed. See Use the analytics dashboard.
Transfer	The agent hands the caller to a person. See Add guardrails.
Outbound call	A call the agent makes, started from your system through the API. See Start an outbound call.
Do-not-call list	Numbers the agent must never call. See Keep a do-not-call list.
Caller memory	What the agent remembers about a caller from earlier calls. See Caller memory.
Test inputs	Saved variable values and Mocked tools for test calls and chats. See Use test inputs.
Mocked tool	A tool that returns a fixed answer in tests instead of calling your API. See Test the agent with mocked tools.
Simulated caller	An AI that plays the caller in a test call. A scenario is a saved simulated call you can run again. See Test with simulated callers.

Workspace and billing

Term	Meaning
Organisation	Your workspace: its agents, numbers, members and settings. Everything you do is inside one organisation. See Invite members and change roles.
Owner, Admin, Builder, Viewer	The workspace roles. See Invite members and change roles.
360 staff access	Time-limited access you give 360 staff to help you. They show as 360 CSM. See Staff access and the audit log.
Audit log	Who changed what in your workspace, and when. See Read the audit log.
Provider key	Your own key for an AI or voice provider, so calls use your account. Added in Settings → AI & voice providers. See Use your own provider keys.
360 CTI	One of the ways to run your phone line, set up by 360. CTI calls bring call details the agent can use as variables, and can transfer to CTI queues. See Connect your telephony.
Prepaid, Postpaid	How you pay: from a balance you top up, or by monthly bill with a budget. See Add credit and read your bill.
Welcome credit	The credit a new organisation starts with, to try the platform. Only the first organisation a person creates gets it. See Your prepaid balance.
Low-balance alert	The balance at which you get an alert, and the wallet.low webhook. See Change the low-balance alert.
Usage limits	The daily allowance for test calls, test chats, simulated calls and AI drafts. See Understand usage limits.

Developers

Term	Meaning
API key	A secret that lets your own systems use the API, with the permissions Read, Write or Calls. See API quick start.
Webhook	A signed request 360 sends to your endpoint when something happens, such as a call ending. See Event webhooks.
Signing secret	The whsec_… secret you use to check that a webhook came from 360. See Check the signature.
Idempotency key	A unique value you send with a request, so sending it twice does it only once. See Idempotency keys.
ETag	A value that says which state of the draft or agent you read. Send it back in If-Match so you don’t overwrite someone else’s change. See ETags and If-Match.