Build an AI Agent Using Local Deepseek Without Coding

Unless you’ve been living under a rock, you’ve probably heard of ChatGPT or AI. In early 2025, a new player entered the scene—Deepseek. Not only is Deepseek a powerful model comparable to OpenAI’s, but it’s also open-source. In this post, I’ll show you how to build a local AI agent using Deepseek—no coding required!

What You’ll Need #

Ollama #

Ollama is a free, open-source platform that allows you to host the Deepseek model. You can install it on your local machine. Open https://ollama.com/download and download the latest version of Ollama for your operating system.

Once set, you will need to download the Deepseek model. You can find the latest version at https://ollama.com/library/deepseek-r1.

If you don’t have a high-end GPU, I recommend downloading the smallest model available. You can always upgrade later.

Run the following command to pull the model:

ollama pull deepseek-r1:1.5b

docker #

Docker is a platform that allows you to run applications in containers. We’ll use it to run n8n.

• MacOS
• Windows
• Linux

Note: If you’re using windows, you may need to use WSL2 to run Docker. You can find instructions here.

n8n #

n8n is a workflow automation tool that lets you connect various services and APIs to automate tasks.

In this post, we’ll use n8n to connect to Ollama and interact with the Deepseek model.

Installing n8n with Docker #

docker volume create n8n_data

docker run -it --rm --name n8n -p 5678:5678 -v n8n_data:/home/node/.n8n docker.n8n.io/n8nio/n8n

Once n8n is running, open http://localhost:5678 in your browser. You may need to create an account to access the n8n interface.

Setting Up n8n #

Credentials #

Before we can start using n8n, we need to set up the Ollama node. This allows us to interact with the Deepseek model.

Search for Ollama in the search bar.

If you follow this guide, the Base URL should be http://localhost:11434. Save the credentials.

Workflow #

Now that our credentials are set up, we can create a workflow. In this example, we’ll build a simple AI agent that generates a Chinese learning story.

from your n8n dashboard, click on the Create workflow button to create a new workflow.

Steps to Create the Workflow #

1. Open the n8n dashboard and click on Create workflow.
2. I’ve created a pre-made workflow that generates a Chinese learning story. 👉 Download the JSON file here

This workflow will automatically run everyday at 9PM. After that it will trigger the AI agent to generate a chinese learning story.

This is the prompt i used to generate the story:

write me a short story written in Chinese to help me with my Chinese practice. Include hanzi, then a pinyin version, then the English translation

example output: 
## 汉字

## Pin yin Version

## English Translation

Since Deepseek is a thinking model, its response may include a reasoning process wrapped between <think> and </think> tags. The workflow automatically removes these tags before returning the final story.

This is the result of the workflow:

## 汉字  
一天，一个小女孩穿着红色的围巾走在街上。她看到一位老人在公园门口摔倒了。小女孩立刻跑过去，扶起了老人。  

## Pinyin Version  
Yī tiān, yī gè xiǎo nǚhái chuān zhe hóngsè de wéijīn zǒu zài jiēshàng...  

## English Translation  
One day, a little girl wearing a red scarf was walking down the street. She saw an old man fall at the park entrance. The little girl ran over and helped him up...  

Conclusion #

And that’s it! You’ve successfully set up a local AI agent using Deepseek and n8n.

Now, you can expand this agent by:

• ✅ Sending the story via email 📧
• ✅ Posting it to Telegram 💬
• ✅ Creating a website to display the stories 🌐

The possibilities are endless!

This is actually my personal AI agent, and I use it to generate stories for my Chinese practice. I hope you find it useful too!

You can check out my workflow’s output here 👉 wendyliga.com/chinese