Beginner Glossary
Use this glossary as a quick reference when you encounter technical terms in the dashboard or documentation.
AI & models
| Term |
What it means |
| LLM (Large Language Model) |
The AI model that reads your question, retrieves relevant knowledge, and generates a response. Examples: Claude, GPT, Gemini. |
| OpenRouter |
A gateway service that gives you access to 40+ AI models with a single API key. FRENZY.BOT uses OpenRouter as its AI provider. |
| RAG (Retrieval-Augmented Generation) |
The technique FRENZY.BOT uses: it retrieves relevant documents from your knowledge base, then sends them to the AI model alongside the user's question. This keeps answers grounded in your data. |
| System prompt |
Global instructions that shape every AI response — personality, tone, rules, and restrictions. Configured in Settings → Model Behavior. |
| Temperature |
Controls how creative or deterministic the AI's responses are. Low = factual and consistent. High = varied and creative. |
| Top-P |
Another creativity control that works alongside temperature. Lower values produce more focused responses. |
| Max tokens |
The maximum number of words/pieces the AI can use in a single response. Higher = longer answers, more cost. |
| Context window |
How much text the AI model can process at once (question + knowledge + conversation history). Measured in tokens. Bigger = handles longer documents. |
| ZDR (Zero Data Retention) |
A privacy mode where the AI provider processes your data but never stores it or uses it for training. |
Knowledge base
| Term |
What it means |
| Knowledge base |
The collection of files, URLs, and FAQs that your bot uses to answer questions. This is what makes your bot specific to your business. |
| Vector database (Qdrant) |
A specialized database that stores text as numerical representations (embeddings) so content can be searched by meaning, not just exact keywords. |
| Embedding |
A numerical representation of text. Similar concepts have similar numbers, allowing the system to find relevant content even when the user's wording doesn't match exactly. |
| Indexing |
The process of converting your files, URLs, and FAQs into embeddings and storing them in Qdrant so they're searchable. |
| Sync job |
A background task that processes and indexes your knowledge base content. You can monitor status in the Sync Jobs page. |
| Term |
What it means |
| Bot |
Your AI assistant instance. Each bot has its own knowledge base, settings, conversations, and widget. |
| Bot ID |
The unique slug identifier for your bot (e.g., sales-assistant). Used in embed code and API calls. |
| Widget |
The chat interface that appears on your website — the bubble in the corner that opens into a chat window. |
| Shadow DOM |
A browser technology that isolates the widget's styles from your website's CSS, preventing visual conflicts. |
| Workspace |
Another name for a bot context. In a multi-bot setup, you switch workspaces to manage different bots. |
Users & security
| Term |
What it means |
| RBAC (Role-Based Access Control) |
A permission system where each user is assigned a role (Admin, Manager, Staff), and each role determines which dashboard modules they can access. |
| OAuth |
A secure login flow where you authorize FRENZY.BOT to access a third-party service (like OpenRouter) on your behalf, without sharing your password. |
| Session |
In the context of dashboard login: how long you stay authenticated before needing to log in again. In the context of chat: a single conversation between a user and your bot. |
Leads & integrations
| Term |
What it means |
| Lead |
A contact captured during a chat conversation — typically name, email, phone, and company. Leads are stored in the database and can be routed to external systems. |
| Integration |
A connector that automatically sends lead data to external tools like Email, Webhook, Google Sheets, or MailChimp. |
| Handoff |
The process of transferring a conversation from the AI to a human agent when the bot can't handle the question or the user requests human help. |
| Lead pipeline |
The status flow a lead goes through: New → Contacted → Qualified → Converted (or Lost). |
Channels
| Term |
What it means |
| Channel |
A communication platform where your bot is available — website widget, WhatsApp, Telegram, Discord, Instagram, Messenger, etc. |
| WAHA |
WhatsApp HTTP API — the Docker-based service that connects FRENZY.BOT to WhatsApp for automated messaging. |
| Webhook |
A URL endpoint that receives data when an event happens (e.g., a new lead is captured). Used for custom integrations. |
| Channel reply style |
Per-channel rules that control the AI's tone and length for different platforms (e.g., shorter replies on WhatsApp, more detailed on the website widget). |
Other
| Term |
What it means |
| Module |
An optional capability you can enable or disable — web search, handoff, text-to-speech, appointment booking, etc. Managed via the Modules page. |