Getting started
Key concepts
The building blocks of Chatisto: workspaces, conversations, the AI agent, scenarios, handoff, operators and contacts, and how they fit together.
Workspace
One website or brand. Each workspace has its own widget (identified by its public key), AI agent and knowledge base, scenarios, inbox, contacts, team and subscription. If you run two websites, create two workspaces and switch between them in the app.
Conversation
A chat with one visitor, or an email thread. Each conversation is:
- Open or closed.
- Handled by the chatbot (AI and scenarios) or in live chat (a human operator).
- Optionally assigned to one operator.
A returning visitor continues their previous conversation if it started less than an hour ago. After that, a new conversation starts.
AI agent
The AI that answers visitors. It answers from your knowledge base (your website, files and text), follows your instructions and skills, and can save a visitor's email or hand off to a human.
Scenario
The flow a conversation follows: a tree of steps such as a welcome message, a button menu, a question, an AI step or a handoff step. Every workspace has one live web scenario and optionally an email scenario. The AI agent itself runs as a step inside the scenario. See Scenarios.
Handoff
The moment a conversation moves from the chatbot to a human. It's triggered by the visitor ("Talk to a human"), by the AI, or by a scenario step. Your handoff settings decide who gets it and what happens when no one is available.
Operator
A team member who answers conversations. Operators are Online (they receive chats) or Away (they see requests, but no sounds and no new assignments). Workspace owners can also change settings and chatbot flows.
Contact
A visitor or email sender. A contact has properties (email, name, phone, company and any custom fields) and their conversation history. Properties are filled by the visitor, the AI, scenario questions, your website's JavaScript API or your team.
AI credits
AI replies use credits from your plan's monthly allowance. Faster models use fewer credits per reply. See Plans & billing.