Types of Chatbots: 7 Main Categories Explained (2026)

Key takeaways:

A chatbot is software that answers questions through text or voice by mimicking a human conversation. When people ask about the types of chatbots, they’re really asking two things at once: how the bot is built, and what job it does.

This guide covers both. You’ll get the short answer first: two broad categories that break down into seven main types. Then a plain-English look at each one, a comparison table, and a fast way to pick the right fit for your business.

How many types of chatbots are there?

There are 2 broad types of chatbots—rule-based and AI—which break down into 7 main categories. Rule-based bots follow pre-set paths; AI bots understand and generate language.

The 7 types are:

1. Menu / button-based—users tap buttons instead of typing.
2. Rule-based (keyword)—follows if-then rules and decision trees.
3. NLP—interprets typed questions in natural language.
4. Machine-learning—improves its responses from past conversations.
5. Generative AI (LLM)—writes original, human-like replies on the fly.
6. Hybrid—combines rule-based control with AI flexibility.
7. Voice bots—handle spoken requests via speech recognition.

Most business chatbots today are hybrid or generative AI bots, which blend the reliability of rules with the natural conversation of large language models.

Chatbot classification: the 2 broad categories

Every chatbot falls into one of two broad classes. Rule-based chatbots follow pre-written scripts and decision trees, so they only answer what they’ve been programmed to answer. AI chatbots use natural language processing and machine learning to understand intent, handle wording they haven’t seen before, and get better the more they’re used.

Rule-based chatbots give users a set of options to pick from. A visitor clicks a category and gets the matching answer. These bots shine on simple, repeatable tasks like answering FAQs, but they stall the moment a question falls outside the script.

AI chatbots use AI, natural language processing (NLP), and machine learning to read what a customer actually means, not just which keyword they typed. They train over time, so their answers grow more accurate and more personal with every conversation.

There’s also a third option that bridges the two: the hybrid chatbot. Hybrids handle common, repetitive queries with scripted speed, then lean on AI for questions that need judgment—and hand off to a human agent when a conversation gets tricky. That mix is why most business bots today are hybrids.

The 7 types of chatbots (by how they work)

Sorted by the technology under the hood, there are seven types of chatbots. They run from the simplest (menu-based bots that show clickable options) all the way to generative AI agents that write original replies in real time. Here’s how they compare before we break each one down.

Chatbot type	How it works	Best for	Conversation ability	Setup effort	Example
Menu / button-based	Users pick from preset buttons or menus—no free typing.	Simple, predictable tasks (store hours, order tracking)	Lowest—fixed paths only	Lowest	FAQ menu bot, IVR-style website widget
Rule-based (keyword)	Matches keywords to if-then rules and decision trees.	FAQs and lead qualification with known questions	Low—breaks on unexpected phrasing	Low	Scripted support/qualification bot
NLP	Uses natural language processing to interpret intent behind typed questions.	Handling varied phrasing of the same question	Medium—understands intent, replies are still templated	Medium	Intent-based support assistant
Machine-learning	Learns from past conversations to improve responses over time.	Use cases that evolve as data grows	Medium–high—adapts with usage	Medium–high	Recommendation / triage bot
Generative AI (LLM)	A large language model generates original, human-like answers in real time.	Open-ended support and natural conversation at scale	Highest—fluent, context-aware, handles the unexpected	Medium (no-code tools) to high (custom)	Tidio Lyro, ChatGPT-style assistants
Hybrid	Combines rule-based control with AI/LLM flexibility—rules for critical paths, AI for the rest.	Most business use cases needing both reliability and nuance	High—guided where it matters, flexible elsewhere	Medium	Support bot with scripted handoff + AI answers
Voice bots	Converts speech to text, processes it, and replies with synthesized voice.	Hands-free and phone-based interactions	Varies—depends on the underlying engine (rule-based to generative)	Medium–high	Alexa, Google Assistant, voice IVR

One note on voice bots: voice is really a delivery channel layered on any engine, from rule-based to generative, which is why their conversation ability varies. We list them separately because the speech interface changes how you build and use them.

Learn more:

• 15 Best AI Chatbots
• 10 Best AI Chatbot Examples from Real Brands
• Chatbot Templates: Free Conversation Flows, Scripts & UI
• 10 Best Chatbot APIs for Websites

1. Menu / button-based chatbots

Menu-based chatbots guide people through clickable buttons instead of free typing. They use decision-tree logic, showing a list of options so a user can tap the one that matches their question.

They’re great for simple FAQs like tracking an order, checking prices, or finding a discount. They’re weaker when someone needs a detailed, off-script answer. If you run an online store, a button bot covers the questions you get over and over.

2. Rule-based (keyword) chatbots

Rule-based chatbots, also called keyword or linguistic bots, run on simple if-then logic: a set keyword triggers a set reply. They’re a strong fit when you already know which questions customers tend to ask.

You can fine-tune the triggers with synonyms and word order so the match is precise. For example, a pricing reply might fire on any of these phrasings:

• “How much does your product cost?”
• “What is your pricing?”
• “What are the prices available?”

The catch: if a customer phrases things in a way you didn’t script, the bot misses. That’s exactly the gap the next type closes.

3. NLP chatbots

NLP chatbots use natural language processing to read the intent behind a full sentence, not just a single keyword. That lets them understand the whole context of a request and answer it properly.

Here’s the difference in practice. Ask a rule-based bot about shipping and it replies to the “shipping” keyword. Ask “What’s the shipping rate for FedEx?” and it still only sees “shipping.” An NLP chatbot catches the FedEx part and answers the actual question.

4. Machine-learning / contextual chatbots

Machine-learning chatbots, sometimes called contextual bots, remember past conversations and improve with every interaction. The more they talk to your customers, the sharper they get.

A contextual bot can store details from earlier chats—a shipping address, billing info, a size preference. If a returning customer wants to reorder, the bot can simply ask whether to use the saved details. That memory makes customers feel recognized, which helps with customer loyalty.

Read more: See how to train an AI chatbot so it keeps getting better.

5. Generative AI chatbots (LLM-powered)

Generative AI chatbots are powered by large language models (LLMs) that generate original responses word by word, rather than picking from pre-written replies. Unlike scripted NLP bots, they can answer questions they were never explicitly programmed for—which makes them the most capable type, but also one that needs guardrails to keep answers accurate and on-brand. Tools like Tidio’s Lyro use this approach to resolve common support questions automatically.

In practice, a generative bot reads your help center, product pages, and policies, then composes a reply that fits the exact question, often across several turns of conversation. Adoption has followed: Salesforce reports AI already resolves around 30% of service cases, on track for 50% by 2027^[1].

Those guardrails matter. Left unchecked, an LLM can “hallucinate” and invent an answer, so the bot has to stay tied to your own content. Lyro, Tidio’s AI agent, works this way: it learns only from the support content you give it, runs on Claude (Anthropic’s model), and creates a ticket for a human when a question falls outside what it knows.

That grounding is what makes the numbers hold up. Lyro automates up to 67% of common customer questions on average. Security firm Cove used it to lift self-service resolution by 70% and cut response times by 80%. The frontier from here is agentic AI—bots that don’t just answer but take actions, like updating an order or qualifying a lead.

[Image note: new landscape asset needed—Lyro / generative AI chatbot in action. Also update the old overview image (3-3.png), which still lists only 6 types; add Generative AI to make 7.]

Read more: Explore the best AI chatbots, or read the full Lyro review for features and pricing.

6. Hybrid chatbots

Hybrid chatbots combine a rule-based layer with an AI layer, so they can run fast scripted flows and still handle open-ended questions. For most businesses, this is the practical default rather than picking one extreme.

Hybrids also pair the bot with live chat. The bot manages the conversation until a request gets complex, then a support agent steps in without breaking the flow. Your customer never has to repeat themselves.

Read more: Compare chatbot vs live chat to see which mix fits your team.

7. Voice bots / voice assistants

Voice bots are a type of conversational AI that works through speech instead of text. They use speech-to-text and text-to-speech, plus natural language understanding, to take a spoken question and answer out loud.

The appeal is hands-free convenience—a person just asks, no typing. The audio is converted to text, the bot finds the answer, and it replies by voice within seconds. Voice bots usually split into hybrid bots (text and voice) and voice-only bots.

Read more: See how chatbots compare to virtual assistants.

Types of chatbots by use case (what they’re used for)

The other way to classify chatbots is by job, not technology. The same AI engine can run as a support bot, a sales bot, or an FAQ bot—what changes is the goal and the content behind it. Here are the most common use-case types, each with a deeper guide.

• FAQ/information bots—answer common questions, share hours, build a knowledge base. See the FAQ chatbot guide.
• Customer support chatbots—automate repetitive tickets and route the rest to agents. See customer service chatbots.
• Sales & lead-generation bots—recommend products, capture emails, qualify leads. See sales chatbots.
• Voice assistants—Alexa, Siri, and Google Assistant carry out spoken tasks.
• Industry bots—tuned for a vertical, like restaurants, real estate, or finance.

Looking for specific tools instead of categories? See our roundup of the best AI chatbots and real chatbot examples. Want to build one? Start with how to create a chatbot for a website.

Rule-based vs AI chatbots: which type should you choose?

Choose a rule-based chatbot if your questions are few, predictable, and rarely change—it’s cheap and fast to set up. Choose an AI chatbot if you field varied questions at volume and want answers that improve over time. Most teams land somewhere in the middle, with a hybrid: scripted flows for transactions, AI for everything else.

	Rule-based	AI chatbot
Setup	Manual scripting	Learns from your existing content
Cost	Low	Low per chat, scales with volume
Flexibility	Only what’s scripted	Handles new phrasing
Accuracy	High on known questions, fails on the rest	High and improving; needs guardrails
Maintenance	Manual updates	Retrains as your content grows

The economics tend to favor AI as you grow. An automated interaction runs roughly $0.50 versus several dollars for a human-handled ticket, and an AI agent answers around the clock. If you’re a small team with a tight FAQ, a rule-based or hybrid bot is plenty. If support volume is climbing, a grounded generative bot like Lyro pays off faster.

FAQ