# Core Concepts

## Core Concepts

<table><thead><tr><th width="228.6171875">Concept</th><th>Description</th></tr></thead><tbody><tr><td>Agent</td><td>A virtual AI assistant that takes and makes calls, or speaks inside your app</td></tr><tr><td>Conversational Pathway</td><td>The visual builder to define how an agent behaves</td></tr><tr><td>Node</td><td>A block inside the pathway (e.g., ask question, transfer, end)</td></tr><tr><td>Variable</td><td>Extracted data from user speech (e.g., name, language)</td></tr><tr><td>DQ Mode</td><td>Locked config with pre-optimized Arabic/English/Dutch stack</td></tr><tr><td>Custom Mode</td><td>Open config mode for selecting any STT, LLM, or TTS provider</td></tr><tr><td>Call Logs</td><td>Complete records and transcripts of every conversation</td></tr><tr><td>Provider Keys</td><td>Where to bring your own OpenAI, Deepgram, etc. credentials</td></tr><tr><td>Phonebook</td><td>Bulk contact lists and campaign launcher</td></tr><tr><td>CallStudio</td><td>Upload historical calls and get AI-powered analysis</td></tr></tbody></table>

**Note on DQ Mode:** When DQ Mode is enabled, VoiceHub automatically selects optimized Arabic/English/Dutch models for speech-to-text, language understanding, and text-to-speech - improving response quality and sentiment accuracy in MENA use cases. Model selection is locked in this mode.