Transcript for:
Automated Reddit Monitoring with AI

In less than 3 minutes, we will make a program that uses AI to monitor whenever someone makes a post on Reddit mentioning your brand. For each post, the program will tell us if it's a complaint, a review, or a recommendation. It will also classify the sentiment of the post as positive, negative, or neutral. And all of this without writing a single line of code. First, go to N8N website and create an account. Once you are in your account, create a new workflow. The way that NA8 works is by chaining together different nodes and each of them have different functionalities. First, we need a trigger. For now, we will just add a manual trigger to run the workflow whenever we click on the note. Next, we will add an HTTP request note. This note let us retrieve a result from a URL. Add the following URL. You can replace the query with your brand and the time parameter can be set to get results from the last hour, day, week, month, or even year. Click on the test button and we get the HTML from the URL we just requested. Next, we will add a markdown note. This note let us convert from HTML to markdown which is much easier for the AI to read. Click on the test button and now we get the markdown version of the HTML. Now we will add a note to interact with chat GPD. Search for OpenAI. Then select the option that says message a model. This will open a new window to configure the open AI node. The first thing we need to do is to set up the credentials. Select the GPT4 mini model from the dropown. I find this to be the most cost effective right now. Now we will send a system prompt to tell the model his role which is to be a helpful intelligent web scraping assistant. Next, we will give it the task of processing the website content and convert it into structured data. So the prompt goes like this. Your task is to take the row markdown of a website and convert it into structured data. If there are no posts, just reply with none. Use the following format. In this case, I am requesting the results to be in HTML format. And the reason is because it's more email friendly. Don't worry, I will paste the prompts in the video description. Finally, in the next box, we just provide the markdown of the website. Next, we will add a send email note. Once our account is configured, we will add a from email and we will add all the emails that we want the data to be delivered to. We add a subject and finally we just drag and drop the email content from the previous note. Click on the test button and surely you will receive an email with all the Reddit post mentioning your brand. As you can see, we retrieved many posts and the AI did a good job classifying the type of post as well as the sentiment. We will add a schedule trigger. So this workflow can run every Monday and we can receive this information at noon. And we are done. In less than three minutes, we made a workflow that uses AI to monitor Reddit for whenever someone mentions our