This page defines the words you will see across the Ori console and the rest of this manual. The definitions are written for day-to-day operations work, not for engineers. Where a term has its own page, the definition links to it.
Terms are grouped by what part of the platform they belong to. Use your browser’s find (Ctrl/Cmd + F) to jump to a word.
At a high level: you build a bot, point a campaign of leads at it, the dialler places calls over a carrier, each answered call runs on a worker in the fleet, and after the call the platform produces a transcript, a recording, a disposition, and optional post-call analysis and QC scores.
You do not need to understand every term to run calls. The ones you will touch most are bot, prompt, campaign, lead, disposition, and callback. The rest are mostly configured once by an administrator.
Bots and prompts
| Term | What it means |
|---|
| Bot | A reusable voice agent configuration: its prompt, voice, language, telephony settings, tools, and post-call rules. One bot can be used by many campaigns and inbound numbers. See Create a bot. |
| Prompt | The written instructions that tell the bot who it is, how to speak, and what to do. The prompt is split into a static part (policy and rules that rarely change) and a dynamic part (per-call values like the customer’s name). See Prompts and variables. |
| System prompt | The main body of the prompt — the bot’s role, rules, and conversation guidance. |
| Opening message | The first thing the bot says when the call connects, before the customer speaks. |
| Variable | A placeholder like {{CUSTOMERNAME}} in a prompt or message that the platform fills in at call time with real data from the lead, the CRM, or the call. See Post-call variables. |
| Policy | A pre-built prompt template for a specific use case (for example a collections or insurance-renewal flow). Selecting a policy gives the bot a tested structure so you supply only the changing details. |
| Tool | An action the bot can take mid-call beyond talking — for example ending the call, transferring to a human, or searching the knowledge base. Tools are configured per bot. |
| Knowledge base | A set of documents you attach to a bot so it can look up answers during a call instead of relying only on the prompt. See Knowledge bases. |
The voice pipeline
These are the engines that turn a phone call into a conversation. They are configured per bot on the voice pipeline screen. See Voice pipeline.
| Term | What it means |
|---|
| STT (Speech-to-Text) engine | Converts what the customer says into text the bot can read. Also called the transcription engine. |
| LLM (Language model) | The “brain” that reads the conversation so far and decides what the bot should say next, following the prompt. |
| TTS (Text-to-Speech) / Voice engine | Converts the bot’s text reply into spoken audio. The chosen voice determines how the bot sounds. |
| VAD (Voice Activity Detection) | Detects when the customer is speaking versus silent, so the bot knows when to listen and when it is safe to talk. |
| Turn detection | Decides when the customer has finished their turn so the bot can respond without cutting them off or waiting too long. |
| Barge-in / interruption | When the customer starts speaking while the bot is talking and the bot stops to listen. |
Calls and outcomes
| Term | What it means |
|---|
| Call | A single phone conversation (inbound or outbound) handled by a bot, with its own record in Call Logs. See Call detail. |
| Inbound call | A call where the customer dials your number and reaches a bot. |
| Outbound call | A call the platform places to a customer, usually as part of a campaign or a CRM-triggered dialout. |
| Transcript | The written, turn-by-turn record of what the bot and customer said, with timestamps. |
| Recording | The audio file of the call. The platform stores it and can share a stable link with your CRM. |
| Disposition | The business outcome of the call in your own categories — for example Connected, RNR (rang, no reply), Not Connected, or Failed. See Dispositions. |
| Sub-disposition | A more specific outcome within a disposition — for example, under a Connected disposition you might record Interested or Not interested. |
| Auto-disposition | A disposition the platform sets automatically by rule when a call clearly needs no analysis — for example a voicemail, a timeout, or a call where the customer never spoke. |
| Post-call analysis | An after-the-call step where a language model reads the transcript and extracts structured outcomes (disposition, summary, promise-to-pay date, and so on) for connected calls. |
| QC (Quality Check) | An optional after-the-call scoring step that rates the conversation on things like empathy and compliance, used for quality monitoring. |
| AHT (Average Handling Time) | The average length of connected calls, shown on dashboards and reports. |
| Connect rate | The share of attempted calls that were actually answered and held a conversation. |
Disposition vs. disconnect reason. The disposition is your business outcome (what to do with this lead). The disconnect reason is the technical cause the call ended (bot, customer, voicemail, RNR, timeout, error). One disposition can come from several disconnect reasons.
Campaigns and dialling
| Term | What it means |
|---|
| Campaign | A list of leads dialled by a chosen bot, with rules for pacing, retries, and working hours. |
| Lead | One contact to be called — a phone number plus any data columns you uploaded (name, amount due, account id, and so on). |
| Dialler | The part of the backend that decides which leads to call and when, places the outbound calls, and hands each answered call to a worker. |
| Pacing | How fast the dialler places new calls. It dials ahead of free capacity so a worker is ready the moment a customer answers, and slows down if too many calls are abandoned. See Pacing and retries. |
| Concurrency / seats | The maximum number of conversations a campaign may have happening at the same time. Ringing calls do not count — only connected conversations. This protects the fleet from being overloaded. |
| CPS (Calls Per Second) | The maximum number of new outbound calls allowed per second, set per carrier to stay within what the telephony provider permits. Separate from concurrency. |
| Retry | A re-attempt of a lead that did not connect (busy, no answer, failed), placed after a configured delay. Every lead is attempted once before any retries run. |
| Callback | A call placed because the customer asked to be called back at a specific time. Callbacks are dialled before fresh leads and retries. See Callbacks. |
| AMD (Answering Machine Detection) | Detecting whether a human or a voicemail/answering machine picked up, so the bot can leave a message or hang up instead of talking to a machine. See AMD. |
| Abandon rate | The share of answered calls with no free worker to take them. A high abandon rate makes the dialler slow down automatically. |
| Working hours | The time window in which a campaign is allowed to dial. Calls outside the window are held until the window reopens. |
Telephony
| Term | What it means |
|---|
| Carrier | The telephony provider that actually carries your calls over the phone network. Carriers hold the dialling limits (such as CPS) and the trunk settings. See Carriers. |
| Trunk | The connection between the platform and a carrier that calls travel through — an inbound trunk receives calls, an outbound trunk places them. |
| DID (Direct Inward Dialling number) | A phone number customers can dial to reach a bot, or the number shown to customers on outbound calls. |
| SIP | The standard protocol used to set up phone calls over the internet between the platform and carriers. You will see SIP referenced in carrier and trunk settings. |
| Exotel | A supported telephony integration where calls reach a bot through an Exotel flow. See Exotel. |
| LiveKit | The media transport the platform uses to carry SIP call audio to and from the bot. It appears in telephony and fleet contexts. |
Fleet and access
| Term | What it means |
|---|
| Fleet | The set of servers that run live calls. The Fleet page shows how much capacity is free across them. See Fleet. |
| Worker | A single call slot within the fleet. Each worker handles exactly one live call at a time; total free workers is your live capacity. |
| CRM | Your customer system of record. The platform can receive dialout requests from it and push call outcomes back to it after each call. |
| Dialout API | The interface a CRM uses to trigger an outbound call to a single number without logging into the console. |
| API key | A secret credential that lets an external system (such as a CRM) call the platform’s API. Shown in full only once, when created. |
| Webhook / post-call push | A message the platform sends to your CRM after each call with the call’s outcome, disposition, recording link, and analysis. See Post-call webhook. |
| Super admin | The highest role. Some areas — Carriers and Settings — are visible only to super admins. |
A note on the names you will see
The console shows brand and feature names, but the underlying service providers (the speech, language, and voice engines) are referred to generically in this manual as the STT engine, the LLM, and the TTS / voice engine. When you pick a specific engine in a bot’s voice pipeline, choose the option that matches your language and quality needs — the screen guides you. See Voice pipeline.
If a term in another page is unfamiliar, come back here first. New to the console overall? Start with the Console tour.