Content and training
| Term | Meaning | Also called |
|---|---|---|
| Chatbot | One AI assistant with its own content, settings, and conversations. | Bot |
| Content | Everything a chatbot learns from: website pages, files, text snippets, custom responses, videos, and connected apps. | Knowledge, knowledge base, training data |
| Source | One place that content comes from, such as a website, a sitemap, a file, or a connected app. | Data source |
| Page | The unit for content limits. One page is 2,500 characters of text. See Plans and limits. | |
| Text snippet | Free text that you type or paste for the chatbot to learn from. Each chatbot has one text snippet. | Custom content (v0 API custom-content), Text |
| Custom response | A question and the answer you write for it. The chatbot prefers this answer when a visitor asks a question with the same meaning. It can still reword it. | Q&A |
| Sync | The chatbot reads a source and learns its current content. | Train, refresh |
| Resync | Sync content again that the chatbot already learned, to pick up changes. It works even when you have no pages left. The page count changes only if the content got longer or shorter. | Retrain, manual refresh |
| Auto-sync | A resync that runs on a schedule. See Keep your content up to date. | Auto refresh |
| Scan | For a sitemap source: check the sitemap, add new URLs, and remove URLs that are no longer in it. |
Answers
| Term | Meaning | Also called |
|---|---|---|
| Instructions | Rules that tell the chatbot how to answer. You select one set of instructions per chatbot. | Prompts (v0 API prompts, dashboard URL /settings/prompts), custom instructions, system prompt |
| Persona | The tone and style of the answers, such as short and direct or formal. You select one persona per chatbot. | Tone |
| Fallback message | The answer the chatbot gives when it cannot find the information in your content. | |
| Message | One message in a conversation, from the visitor or from the chatbot. Messages are also the unit for the monthly message limit. One question and its answer count as 2 messages. See Plans and limits. | |
| Conversation starter | A suggested question that visitors can select to start a conversation. | Quick prompt (v0 API quick-prompts), suggestion, starter |
| Follow-up suggestion | A suggested next question shown under an answer. You can write your own. The AI can also suggest them (smart follow-up questions). | Follow-up prompt (v0 API follow-up-prompts), followup |
Conversations and human support
| Term | Meaning | Also called |
|---|---|---|
| Conversation | All the messages between one visitor and the chatbot or your team, from start to end. | Thread (v0 API threads), chat |
| AI mode | The chatbot answers the visitor. | |
| Human mode | Your team answers the visitor. The chatbot does not reply. | Agent mode (AGENT in the API) |
| Human support | The features that let visitors reach your team from the chat. See Set up human support. | |
| Escalation | A request to move a conversation to your team. The visitor or the AI can start it. It puts the conversation in human mode and sends notifications. | Handoff |
| Take over | A team member starts answering a conversation from the dashboard. The AI stops replying, and the visitor cannot switch back to AI. No escalation notification is sent. | |
| Release to AI | A team member gives a conversation back to the chatbot after a take over. | Switch to AI |
| Resolve | Mark a conversation as done. The visitor sees that the conversation is closed and can start a new one. | Close, mark as resolved |
Leads
| Term | Meaning | Also called |
|---|---|---|
| Lead | Contact details that a visitor gives in the chat, such as name and email. | |
| Lead form | The form in the chat that collects lead details. | Contact form |
| Pre-chat user details | A form that asks visitors for their details before the chat starts. You set it in Leads Settings. | User data |
Account and team
| Term | Meaning | Also called |
|---|---|---|
| Owner | The account that created a chatbot. The owner’s plan sets the limits for the chatbot. | Creator |
| Team member | A person you invite to work on a chatbot. Each member has a role. See Team members. | Member, seat |
| Role | What a team member can do: Agent, Manager, Admin, or Super Admin. |
Developers
| Term | Meaning | Also called |
|---|---|---|
| API token | A key for the v2 API, the CLI, and MCP, with scopes that limit what it can do. You create it on the Agents page. See Authentication. | Scoped token, v2 token |
| API key (legacy) | The older single key for the v0 and v1 API. It shows on the Billing page on Growth plan and above. Use API tokens for new work. | API key, v0 key |
| JavaScript widget SDK | The window.$sitegpt commands that control the chat widget on your website. See Widget SDK. | SDK methods, $sitegpt |
| TypeScript API SDK | The @sitegpt/sdk package that calls the v2 API from your server code. See TypeScript SDK. | SDK |
| Webhook | A request that SiteGPT sends to your URL when an event happens, such as a new lead. See Webhooks. |