Chatbot
A chatbot is software that answers in a chat window using preset rules or a language model, while an AI employee takes on a role and looks at store data.
A chatbot is software that chats in place of a person, in a chat window or a messaging app. There are two main kinds. Rule-based ones show ready options and follow a path drawn in advance: “type 1 to check an order”. Ones that use a language model try to understand a freely typed question and write a sentence in reply.
Both share the same limit: what the answer rests on. A chatbot knows only what it was given. If it cannot look at the order, the cart or the earlier conversation in your store, it cannot answer the customer's real question. When a question falls outside the flow, it stalls, goes back to the menu or writes a generic line.
Why It Matters
- Knowing what the term covers lets you tell what a tool really does from how it is marketed.
- Building a menu flow and building an employee that looks at customer data are different efforts with different results.
- A customer whose question is not answered the first time drops the conversation, so what the bot rests on decides the outcome.
How It Works
- A flow or a model is chosenA rule-based bot is built as a menu that moves step by step. A bot with a language model interprets a freely typed question.
- The customer writesThe customer picks from a menu or types their own sentence.
- The bot repliesA preset text, a product link or a sentence written by the model comes back.
- It hands over when stuckA well-built bot passes the conversation to the team when it gets stuck. If it does not, the customer stays in a loop.
Example
Liva Butik's site has a menu bot: “1 Order, 2 Return, 3 Product”. Mert K. writes “Where is #1042?”. The bot goes back to the menu and says “Choose an order”, because it cannot look at the order record. Mert ends up writing to Burak Demir. With an AI employee in the same store, it could find the order in the connected store record and write the shipping status. The difference is in the data it looks at.
| Classic chatbot | AI employee | |
|---|---|---|
| Where it finds the answer | Preset menus and texts | Store data and approved rules |
| Does it know the customer? | Mostly no, every session starts over | Looks at orders, cart and past chats |
| Outside the flow | Stalls or returns to the menu | Answers or hands over to the team |
| Scope | A single flow | A role such as sales or support |
Not to Be Confused With
Chatbot is the general name for a type of software; not every tool that chats does the same job. Creato does not call its own product a “chatbot”. It says “AI employee” and “AI assistant”, because the job is more than a menu flow.
In Creato
In Creato, the AI Employees that chat with customers are set up to answer by looking at the store's data.
- AI Employees answer knowing the order, the cart, the past conversation and the store catalogue.
- Before going live, you try it first with a live test.
- Stock and price are checked from the connected store. If there is no connection or no data, stock is not answered as if it were verified.
Frequently Asked Questions
What is the real difference between a chatbot and an AI employee?
When choosing a tool, ask one question: what does it look at when it answers? A tool that pulls replies from uploaded text or a flow is in the chatbot class. A tool that reads orders, carts and past conversations and does the work of a role is an AI employee. See the AI employee page.
What do I need to build a chatbot?
For a rule-based bot, drawing the question and answer flow is enough. For a bot with a language model you also need a source of knowledge: return terms, product descriptions, frequently asked questions. After building it, try it with real customer questions.
Can a chatbot give a wrong answer?
Yes. Whatever the type, the chance of error in an automated reply is not zero. To lower the risk, let changing facts such as stock and price be read from the connected store, and make sure the bot hands over to a person on what it does not know.