Glossary
Chatbot (Chat Bot)
Software that holds a conversation with people over text or voice. In a chat app a bot is an automated participant in the room, with its own name and profile, that reads messages and replies by rules or by AI.
General definition
A Chatbot (Chat Bot) is a program that carries on a conversation with human users through a messaging interface. The word bot in chat simply means an automated account: a participant that is not a person, that reads what is posted and responds or acts by itself. Chatbots range from a simple auto-reply that answers “hours” with opening times to an AI chatbot that writes original answers with a large language model.
Three families cover almost every chatbot in use:
- Rule-based: decision trees, menus and keyword matching with pre-written replies; predictable, cheap and easy to audit, but brittle the moment a user goes off script
- AI (generative): a large language model interprets any wording and composes a reply, often grounded in your documents through retrieval
- Hybrid: rules for sensitive or transactional steps (identity checks, payments, escalation) and a model for open questions, which is how most production bots are built today
The idea is older than the web. ELIZA (1966) mimicked a therapist with pattern matching; SmarterChild (2001) answered millions of AIM and MSN Messenger users from a scripted knowledge base; Siri (2011) brought voice assistants with intent classification to phones; and the release of ChatGPT in late 2022 made generative chatbots the default expectation for what a bot can do.
Inside a chat platform a bot is a participant like any other. It has its own user account, avatar and profile, it joins rooms or channels, it receives messages through the same real-time protocol as people do, and it posts replies, files, buttons or links back into the room. Typical features are greetings and onboarding, FAQ answers, slash commands, form-filling flows, notifications and reminders, translation, moderation, and handover to a human. Building one means choosing a channel (your own app, a website widget, WhatsApp or Slack), a brain (rules, a model, or both), a knowledge source, and the guardrails around it, then connecting it to the chat system through an SDK or API. The broader field is conversational AI.
In the Ethora ecosystem
Ethora started with chat bots. The earliest versions of the platform shipped a bots framework in which a bot was a room participant with its own user profile, listening to messages over XMPP and replying in the same room, and that framework still lives in the open-source monorepo. The same idea now runs on the AI SDK: an AI agent is attached to a room or channel, carries a system prompt, a knowledge base for retrieval and its own model choice (cloud API or a self-hosted model through an OpenAI-compatible endpoint), and can call tools. Several agents can share one room and reply to each other, and a heartbeat scheduler lets a bot speak first.
For teams adding a bot to an existing product, the Chat SDK supplies the rooms, users, files and real-time delivery on web, React Native, iOS and Android, so the bot appears in the same conversation list as human contacts, and the embeddable assistant widget puts the same bot on a website with one script tag. Messages to and from bots pass through Trust and Safety moderation and are kept in the server-side archive with the audit trail. See the AI agent and AI SDK entries for the agent side of the story.
