All chatbots — whether free or paid, rule-based or AI-powered — follow a series of steps when interacting with users. The sophistication of each step determines how “smart” the chatbot feels.

We’ll explore each in detail.

User Input Processing

Every chatbot interaction begins with input detection. Chatbots process one of:

• Text
• Button selection
• Voice command
• Menu choices
• Form data
• Webhook data
• API triggers

The bot first needs to capture and interpret what the user is trying to do.

Examples of input:

• “Where is my order?”
• “Show me red Nike shoes under $80”
• “Book a table at 7pm”

Free chatbot builders like Landbot, ManyChat, and Chatfuel capture button/menu inputs easily.
AI chatbots use NLP engines to process free-form text.

Intent Recognition (NLP + NLU)

After capturing the input, the chatbot must determine the intent — the meaning behind the message.

This uses technologies like:

• NLP (Natural Language Processing)
• NLU (Natural Language Understanding)
• ML Models (Machine Learning)

Example:

User message:

“I need help with my order”

Possible intents:

• Order Tracking
• Refund
• Delivery Delay
• Cancellation

AI chatbots like Tidio AI, Intercom Fin, and ChatGPT Assistants can identify the correct meaning automatically, even if the user types it in their own unique way.

Rule-based bots rely on keyword matching, such as:

• “order”
• “tracking”
• “refund”

This method is simpler but less accurate.

Entity Extraction

Modern chatbots extract specific data elements (called entities) from the message.

Entities could include:

• Dates: “tomorrow”, “next Monday”
• Products: “blue hoodie”, “iPhone 15 case”
• Locations: “New York”, “Dubai”
• Order numbers: “#5420192”
• Categories: “flights”, “insurance”, “shoes”

This is crucial in eCommerce chatbots.

Example:

User: “I want Adidas sneakers in size 9 under $100.”

Entities extracted:

• Brand: Adidas
• Category: Sneakers
• Size: 9
• Price: <100

AI chatbots excel at entity extraction.
Free chatbots can only extract basic keywords.

Response Generation

Once the bot identifies the intent and entities, it generates a response.

There are 3 major types of responses:

1. A) Rule-Based Responses

Predefined responses such as:

• “Your order is on the way!”
• “Here is our return policy.”

Tools: Landbot, Chatfuel, ManyChat (free plan)

1. B) AI-Generated Responses

AI bots generate responses in real-time using large language models.

Tools:

• ChatGPT Assistant
• Tidio AI
• Botsonic
• Intercom Fin
• LiveChat AI

These tools dynamically generate:

• Answers to complex questions
• Personalized replies
• Product suggestions
• Context-aware messages

1. C) API-Based Responses

Chatbots can pull data from external systems:

• Shopify (order tracking, product lists)
• WooCommerce
• CRM systems
• Databases
• Google Sheets
• Payment APIs

Example:

User: “What’s my order status?”
Bot calls Shopify API → returns real-time order info.

Action Execution (Workflow Automation)

Chatbots do MUCH more than reply with text.
They can perform actions.

Examples of chatbot actions:

• Sending an email
• Creating a support ticket
• Booking an appointment
• Processing a refund
• Adding items to cart
• Applying a discount code
• Logging a CRM lead
• Triggering a webhook
• Sending a WhatsApp message

Tools like ManyChat, Make.com, Zapier, Tidio, and Intercom shine in automation.

AI chatbots can also perform complex multi-step actions using “AI Actions” or “Assistant workflows.”