# What is Chatbot?

Also called Chat Bot, Conversational Agent.

A chatbot is a software application that holds a text conversation with a person through a messaging interface. Implementations range from rule based systems that match keywords to scripted replies, through to systems driven by large language models that generate each response. The word describes the interface and product form, not any particular underlying technology.

The earliest chatbots pattern matched input and echoed templated text. Intent based bots improved on that by classifying an utterance into one of a fixed set of intents, extracting entities such as dates or order numbers, and running the matching flow. Model driven bots skip the fixed intent list and produce a reply directly, which widens coverage but makes behavior harder to enumerate in advance.

Because a chatbot occupies a channel people already use, it can absorb repetitive requests at any hour and keep a written record of what was said. Design effort concentrates on the edges rather than the happy path: what happens when a request falls outside scope, how a conversation is escalated, and how plainly the bot states what it cannot do.

Two confusions are common. First, chatbot is often used loosely for any automated messaging, including one way notification senders that never interpret a reply. Second, people assume a chatbot is stateless. Most keep conversation state, and a system that forgets what was said two messages ago reflects a design choice or a context limit rather than a defining property.

Quality is usually measured by resolution rather than volume. Containment, the share of conversations that end without escalation, is easy to report and easy to game, since a bot that offers no exit contains everything. Pairing containment with an outcome signal, such as whether the person came back with the same question, gives a far more honest picture.

## Key points

- Names an interface form, not a specific technology
- Rule based, intent based, and model driven variants all exist
- Escalation path matters as much as answer quality
- Most keep conversation state across messages
- Containment without an outcome measure is a misleading metric

## In practice

A bicycle retailer puts a chatbot on its order tracking page. A customer asks where their wheelset is, the bot reads the order number from the account session, checks the carrier status, and reports that delivery is scheduled for Thursday. When the same customer asks to change the delivery address, the bot recognizes it cannot do that safely and hands the conversation to the support queue.

## Related terms

- [Conversational AI](/en/glossary/conversational-ai)
- [Chat Widget](/en/glossary/chat-widget)
- [Conversation Thread](/en/glossary/conversation-thread)
- [Channel Routing](/en/glossary/channel-routing)
- [Email Agent](/en/glossary/email-agent)

[Back to the AI Glossary](/en/glossary)