--- title: "AI chatbot vs rule-based chatbot vs AI agent: shop examples" description: AI chatbot, rule-based chatbot or AI agent? See how each one handles real messages from Bangladeshi shops, where each fails, and which mix to choose. url: https://salemag.net/blog/ai-chatbot-vs-rule-based-chatbot language: en published: 2025-12-09 updated: 2025-12-09 author: SaleMag Team tags: - ai chatbot - rule-based chatbot - ai agent - chatbot comparison --- # AI chatbot vs rule-based chatbot vs AI agent: shop examples > A rule-based chatbot replies with fixed answers when a message matches a keyword or button. An AI chatbot understands free text and writes answers from your own content. An AI agent is an AI chatbot that can also act, such as checking stock or placing an order. Most shops need rules for fixed answers and AI for the rest. - Rule-based bots are predictable and cheap but break on any wording they were not set up for. - AI chatbots understand Bangla, Banglish and typos, but must be limited to your own content to stay accurate. - AI agents add actions: product lookups, carts, order forms and handover to a person. - The best setup is a mix: keyword rules first for fixed answers, then the AI for free-text questions. - Whatever you choose, plan who answers when the bot cannot. Picture three shops getting the same message at 11 pm: "apu oi pink ta M e ache? Dhakar baire koto din lagbe?" The first shop has a rule-based bot. It sees no keyword it knows and sends the menu: "Press 1 for price, 2 for delivery." The customer closes the chat. The second shop has an AI chatbot. It understands the question, finds the delivery time in the shop's FAQ, but has to say "please check with our team about stock". The third shop has an AI agent. It looks up the pink kurti, sees two M left, answers both questions, and asks if she wants to order. All three are called "chatbots". They are very different tools. Here is how each works, with examples from Bangladeshi shops, and how to pick the right mix. ## What is a rule-based chatbot? A rule-based chatbot is a bot that sends a fixed reply when a message matches a rule: a keyword, a button tap or a menu number. It does not understand language. It checks words against a list. Typical rules on a Facebook page or WhatsApp number: | Customer sends | Rule | Bot replies | |---|---|---| | "price" | Keyword contains "price" | "Please send the product photo, we'll tell you the price." | | First message ever | Welcome rule | "Assalamu alaikum! Welcome to ShopName." | | Any message after 11 pm | Away rule | "We're closed now, we'll reply at 10 am." | | Taps "Delivery info" | Button | "Inside Dhaka ৳60, outside ৳120." | **Where it works:** fixed answers that never change by wording. An offer code, business hours, a bKash number for advance payment, the welcome message. **Where it fails:** everything else. "Dam koto", "koto taka", "prize koto" and "eta koto pore" all mean "price?", and a rule only catches the ones you thought of. Two questions in one message, like the pink kurti example, defeat it completely. ## What is an AI chatbot? An AI chatbot is a bot built on a large language model that understands free-text messages and writes replies, ideally from your own content only. It reads the meaning of a message, not just its keywords. The same messages go very differently: ``` Customer: Dhakar baire delivery charge koto r koidin lage? AI chatbot: Dhakar baire delivery charge ৳120, shadharonoto 2-4 diner moddhe pouche jay. ``` ``` Customer: exchange kora jabe? size choto hoise AI chatbot: Ji, delivery er 3 diner moddhe tag shoho ferot dile size exchange kora jay. Apnar order number ta diben? ``` The AI understood two questions in one message, Banglish spelling and a typo ("hoise"). That is the big step up from rules. **Where it works:** the long tail of questions about delivery, returns, fabric, sizes and policies, in Bangla, English and [Banglish](https://salemag.net/glossary/banglish). **Where it fails:** anything not in its content, and anything that needs live data or an action. A plain AI chatbot reading a price list from last month will quote last month's prices. And if it is allowed to answer from general knowledge, it may make up a policy that sounds right. A good AI chatbot is told to answer only from your knowledge base and to say "I don't know" otherwise. ## What is an AI agent, and how is it different? An [AI agent](https://salemag.net/glossary/ai-agent) is an AI chatbot that can also use tools to take actions: look up a product's live price and stock, add items to a cart, send an order form, or hand the chat to a person. It does not just answer. It does the next step. Here is the pink kurti chat with an agent: ``` Customer: apu oi pink ta M e ache? Dhakar baire koto din lagbe? [agent searches catalog: "pink kurti" -> Pink Printed Kurti, ৳1,250, M: 2 left] Agent: Ji, Pink Printed Kurti M size e 2 ta ache, dam ৳1,250. Dhakar baire 2-4 din lage, charge ৳120. Order korben? Customer: hae korbo [agent builds cart and sends order form] ``` The difference is the bracketed lines. The agent checked live stock instead of guessing, then moved the sale forward. That is why agents matter for chat-based shops: most messages are buying questions, and buying questions need live data. **Where it fails:** judgement calls. A wholesale price for 50 pieces, a complaint about a torn parcel, a customer who wants to change an address on a shipped order. A well-built agent recognises these and hands them over, which is why [human handover](https://salemag.net/glossary/human-handover) is part of any serious setup. ## How do the three types compare side by side? The three types differ in what they understand, what they can do and what they cost to run. The table compares them on the points that matter to a shop. | | Rule-based chatbot | AI chatbot | AI agent | |---|---|---|---| | Understands free text | No, keywords only | Yes | Yes | | Bangla, Banglish, typos | Only words you listed | Yes | Yes | | Answers from | Fixed replies you wrote | Your knowledge base | Knowledge base plus live tools | | Live price and stock | No | Only if the content is current | Yes, looks it up | | Takes an order | Only through rigid menus | No, collects details at most | Yes, builds cart and sends a form | | Risk of wrong answers | Low, but often irrelevant | Medium if allowed to guess | Lower when it must look facts up | | Setup effort | Many rules for good coverage | A good knowledge base | Knowledge base plus clean catalog | | Running cost | Software only | Software plus AI usage | Software plus AI usage | ## Which one should a Bangladeshi shop use? Most shops should use rules and AI together: keyword rules for the handful of fixed answers, and an AI agent for the free-text questions and orders. The two do not compete. They cover different messages. A simple way to decide, by shop type: | Shop situation | Recommended setup | |---|---| | Under 20 chats a day, owner answers | WhatsApp Business app greeting and away messages are enough | | A service business with a long FAQ (clinic, coaching centre) | AI chatbot on a strong knowledge base, with handover | | A Facebook or Instagram shop with 50+ product questions a day | AI agent with the catalog and order taking | | Running a fixed promo code or giveaway | Add a keyword rule for that word, on top of the AI | The order matters. Let fixed rules run first, so "offer" or "11.11" always gets your exact campaign reply, and let the AI handle anything that does not match a rule. If you are still answering "price koto?" by hand, start with our guide to [answering price questions automatically](https://salemag.net/blog/answer-price-koto-messages-automatically). ## How does SaleMag combine rules and AI? SaleMag runs both in one order. Keyword rules, a welcome message and an out-of-hours reply run first, so fixed answers stay fixed. If no rule matches, the [AI chatbot](https://salemag.net/features/ai-chatbot) answers from your knowledge base of files, web pages, FAQs, text and products, in the language the customer writes in. If the customer asks for a person, or the bot cannot answer and "hand over when unsure" is on, the chat goes to your team in the shared inbox. With [AI order taking](https://salemag.net/ai-agents/ai-order-taking) switched on, the same chatbot becomes an agent: it searches your catalog before quoting any price, builds the cart, and sends an order form for a cash-on-delivery order. It runs on your own OpenAI, Anthropic (Claude) or Google Gemini key, so AI usage is billed by that provider. ## Next steps Open your inbox and sort yesterday's messages into three piles: fixed answers, free-text questions and buying questions. The fixed pile becomes keyword rules. The other two tell you whether you need an AI chatbot or a full agent. To try both on your own page or WhatsApp number, [start free with SaleMag](https://salemag.net/register). ## FAQ ### What is a rule-based chatbot? A rule-based chatbot is a bot that replies with pre-written answers when a message matches a keyword, a button or a menu choice. It does not understand language; it only follows the rules you set. ### What is the difference between an AI chatbot and an AI agent? An AI chatbot understands questions and replies from your content. An AI agent also takes actions with tools, such as searching the catalog, building a cart, sending an order form or handing the chat to a person. ### Is an AI chatbot better than a rule-based chatbot? Not always. An AI chatbot handles free text and unusual wording far better, but a rule-based reply is more predictable for fixed answers such as an offer code or business hours. Many shops use both. ### Can a rule-based chatbot understand Banglish? Only for the exact words you add as keywords. A rule for 'price' will miss 'dam koto', 'koto taka' and 'prize koto' unless you add each one, which is why rule-based bots often fail on Banglish. ### Do AI chatbots make mistakes? Yes, an AI chatbot can give wrong answers if it is allowed to guess or its content is out of date. Limiting it to your own knowledge base, making it look up prices, and handing unclear chats to a person keep mistakes rare. ### Which chatbot type costs more to run? A rule-based chatbot usually costs only the software subscription. An AI chatbot or agent also has AI usage costs, billed per use by the AI provider, which grow with the number of messages it answers.