Core concepts
A short glossary of the terms you'll meet across Promptly. Each one links to the page where it's covered in depth.
Tenant / workspace
A tenant is an isolated account — your company's workspace. All your knowledge, settings, conversations, and team members live inside it, and Promptly keeps every tenant's data strictly separated. When you log in, you're working inside one tenant. See Data isolation & storage modes.
Knowledge item
A single unit of content the bot can answer from: a manually written article, an uploaded file, or a crawled web page. Knowledge items carry a title and an optional category. See Adding knowledge.
Embedding
When you add a knowledge item, Promptly converts its text into an embedding — a numeric vector that captures meaning. Embeddings let the bot find content by semantic similarity rather than exact keyword match. See Enrichment & re-indexing.
Retrieval / RAG
Retrieval-Augmented Generation (RAG) is the core loop: for each visitor message, Promptly retrieves the most relevant knowledge, then asks the AI model to answer using only that content. This is what enforces the no-hallucination golden rule. See How retrieval works.
Product catalog
A live connection to your store or product API. Once mapped and synced, the bot can search your products and render them as visual cards in the chat. See Connecting a catalog.
Widget
The embeddable chat interface your visitors use — a script tag you place on your site. You control its look and behavior in the Widget Customizer. See Customizing the widget and Embedding the widget.
Escalation
The handoff from bot to human. When the bot can't help (or the visitor asks for a person), the conversation is routed to your agents in the Inbox. See Escalation settings and Working the Inbox.
Modules
What you subscribe to. Chat is the base module — the assistant, the knowledge base, and the widget. FAQ, Product Catalog, Documentation, and Escalation are added on top as you need them; the last two are billed per seat. A feature you don't have shows a lock naming the module that unlocks it. See Pricing & limits.
Your own AI key (BYOK)
Chat replies run on your AI provider account — OpenAI, Anthropic, Gemini, GLM, or Kimi — not on a shared platform key. You add the key once in Dashboard → Settings → AI Keys, it's stored encrypted, and your provider bills you directly for tokens. Because of that, Promptly doesn't meter conversations: there's no message allowance to run out of. See Model & behavior settings and API keys & encryption.
Trial
Every new workspace gets 14 days with every module unlocked. The trial runs on your own AI key too, and applies lower limits on knowledge base size and crawling. After the trial, a workspace without a subscription becomes read-only until you subscribe — nothing is deleted. See Pricing & limits.
Roles
Team members are assigned roles that control what they can see and do — for example, who can change settings versus who only works conversations. See Roles & permissions (RBAC) and Team & roles.