Build an AI Finance Assistant LINE Bot with n8n, Docker, and RAG (2025)

Introduction to n8n and Building an AI Financial Assistant LINE Bot

What is n8n?

n8n is a popular automation platform that allows users to create automated workflows, reducing the burden of various tasks in life and work. In this course, we will guide you through building an AI financial assistant LINE Bot that can answer questions about stock prices, financial knowledge, and even provide trading advice.

Course Outline

The course is divided into four weeks:

• Week 1: Basic introduction to n8n and environment setup, including an overview of n8n and Docker (a tool for running n8n locally).

• Week 2: Integrating LINE to create a LINE Bot, covering concepts like Messaging API (Push and Reply messages), and using Ngrok to generate a temporary external URL.

• Week 3: Connecting to a stock API and setting up automated processes to push stock information.

• Week 4: Building the final AI financial assistant LINE Bot, involving concepts of Large Language Models (LLM) like ChatGPT and the Retrieval-Augmented Generation (RAG) technique.

n8n Basics

• Pronunciation and Name Origin: n8n is pronounced as "n-eight-n". Its full name is nodemation, with "n" at the beginning and end, and eight letters in between. Similar abbreviations like i18n (Internationalization) are common in English.

• Components of nodemation: It consists of "node" and "mation". "Node" refers to a Node View, where the interface is made up of individual nodes. These nodes can be connected and configured to achieve desired results. It also has a double meaning as it is built using Node.js. "Mation" represents the concept of automation.

Features of n8n

• Low-Code Platform: Allows users to create workflows without extensive coding. However, custom functions can be added through programming.

• Integration Capabilities: Can connect to various apps and services such as Google and OpenAI.

• Convenience: Can be installed locally using npm or Docker.

• Security: Since the n8n server can be hosted locally, data privacy is ensured as it is not uploaded to the cloud.

n8n Pricing

n8n offers different plans with varying capabilities. The cloud versions are paid, with a 14-day free trial. During the trial, users can execute 10,000 workflows and create up to 15 workflows. After the trial, if the plan is not upgraded within 28 days, all projects in the workspace will be automatically deleted.

Comparison with Other Automation Platforms

• Zapier, Make, and n8n: Zapier and Make are more beginner-friendly, while n8n requires some programming knowledge. n8n offers greater flexibility due to its higher technical threshold.

• Cost: Zapier and Make are paid after the free trial, while n8n is also paid after the 14-day trial. However, n8n is open-source, allowing users to deploy it locally for free.

• Integration: Zapier has a large number of integrations (over 6,000), Make has around 1,500, and n8n has approximately 1,000 (but this number is increasing as more developers contribute).

• Workflow Complexity: n8n supports more advanced logic and branching, while Zapier and Make are more basic.

n8n Interface

• WorkFlows: This is where users create and manage their workflows.

• Credentials: Stores API keys and other credentials used in the workflows.

• Executions: Records of past workflow executions.

Types of Nodes in n8n

• Triggers and Actions: Determine when the workflow is triggered.

• Core nodes: These are the main nodes for connecting to APIs, such as LINE and Google.

• Cluster nodes: Allow for nested nodes, like the AI Agent.

• Credentials: Stores API keys and other credentials.

• Community nodes: Developed by other users and available for download.

Docker

• What is Docker? Docker is an open-source container management platform used for developing, running, and deploying applications. It uses containers to package applications, enabling them to run on different operating systems.

• Design Philosophy: Build once, deploy anywhere. It provides a consistent environment from development to deployment.

• Core Concepts:

  • Image: A template or blueprint of an application, including all dependencies.

  • Container: Runs the application image and provides isolation.

  • Volume: Stores data used by the application.

  • Dockerfile: An automated script for building the image.

Docker Architecture

• Docker Client: Sends commands to the Docker Daemon.

• Docker Daemon: Manages containers and images on the Docker Host.

• Docker Registry: A public platform for sharing images. The official registry is Docker Hub.

Difference between Docker and Virtual Machines (VM)

• Resource Allocation: Docker does not require the allocation of physical resources like VMs, so it starts up quickly.

• Resource Usage: Docker uses less resources compared to VMs.

Setting up n8n Locally with Docker

• Using Docker Volume:

  • Create a volume: docker volume create n8n_data

  • Run the container: docker run -p 5678:5678 -v n8n_data:/home/node/.n8n n8nio/n8n

  • Access the service at http://localhost:5678 or http://127.0.0.1:5678.

• Using Docker Compose:

  • Create a docker-compose.yml file with the necessary configuration.

  • Start the container: docker compose up -d

  • Access the service at http://localhost:5678.

Updating the n8n Image

• Using Docker Volume:

  • Stop the container: docker stop n8n

  • Pull the latest image: docker pull n8nio/n8n

  • Run the container: docker run --rm -p 5678:5678 -v n8n_data:/home/node/.n8n n8nio/n8n

• Using Docker Compose:

  • Stop and remove the container: docker compose down

  • Pull the latest image: docker compose pull

  • Start the container: docker compose up -d

Week 2: Integrating LINE with n8n

LINE Messaging API

• Push API: The LINE Bot sends messages to the user.

• Reply API: The user sends a message to the Bot, and the Bot replies. Since Reply API is free, it is preferred in design.

LINE Messaging API Flow

• The user sends a message (event) to the LINE Platform.

• The LINE Platform notifies the Bot Server.

• The Bot Server processes the message and uses the Messaging API to send a reply back to the LINE Platform, which then delivers it to the user.

LINE Official Account Manager vs LINE Developers

• Official Account Manager: Used for managing the official account, does not require programming.

• Developers: Used for using LINE API, requires programming. The official account manager has more limitations but lower development costs, while the developers' platform offers more flexibility at a higher cost.

Provider and Channel

• Provider: Can be an individual or an organization that provides services and can access user information.

• Channel: Allows the Provider to use API services. Different Channels can be created under a Provider.

User ID

• The User ID is unique within the same Provider. Different Providers may have different User IDs for the same user.

Ngrok

• Purpose: Ngrok helps create a temporary external server to solve the problem of local n8n not being accessible from outside. It enables real-time Webhook testing and debugging.

• Working Principle: Ngrok creates a tunnel between the user and the local server, allowing external requests to be routed to the local server.

Setting up Ngrok

• Download and Install: Download the Ngrok executable from the official website.

• Authenticate: Copy and paste the authentication token provided by Ngrok.

• Create a Public URL: Run the command ngrok http 5678 to create a temporary external URL.

• Update n8n Configuration: Modify the docker-compose.yml file to include the Ngrok URL as the Webhook URL.

• Restart n8n: Stop and remove the existing container, then start a new one using docker compose up -d.

Week 4: Introduction to RAG (Retrieval-Augmented Generation)

What is RAG?

RAG is a technique that combines retrieval of relevant information with the generation of responses by a Large Language Model (LLM). It addresses the limitations of LLMs, such as limited knowledge and the inability to update information.

RAG Pipeline

• User Query: The user asks a question.

• Document Chunking: Relevant documents are divided into smaller chunks.

• Vectorization: The chunks are converted into vectors.

• Embeddings: The vectors are stored in a vector database.

• Retrieval: The most relevant vectors are retrieved based on the user's question.

• Augmentation: The retrieved information is combined with the user's question.

• Generation: The combined information is sent to the LLM to generate a response.

Comparison between RAG and Direct LLM Use

• Knowledge Source: Direct LLM use relies on pre-trained data, while RAG retrieves additional relevant information.

• Updating Information: RAG allows for easier updating of information by simply updating the documents, while LLMs require retraining.

• Source Awareness: RAG can identify the source of the information, while LLMs may not.

• Accuracy and Hallucinations: RAG provides more accurate responses based on retrieved knowledge, reducing hallucinations.

• Speed: RAG is slower due to the additional retrieval step.

• Use Cases: Direct LLM use is suitable for general conversations, while RAG is more appropriate for specialized tasks.