Follow ZDNET: Add us as a preferred source on Google.
ZDNET’s key takeaways
- YouTube stopped comment emails, hurting engagement workflows.
- Gemini helped rebuild alerts using a simple Python script.
- AI makes one-off automation projects fast and practical.
Comments are the lifeblood of YouTube video engagement. Comments help creators engage their audience. They are also a signal to The Algorithm that viewers are engaged.
YouTube prioritizes engaged viewers, so the more comments, the better. One of the best ways to ensure more regular comments is for the video producer to respond to readers quickly and clearly.
Also: 7 AI coding techniques I use to ship real, reliable products – fast
Until the middle of last year, YouTube sent out an email whenever a viewer commented on a video. That was my trigger. If I received an email saying that I had a comment, I clicked in and responded.
But at the end of June, YouTube quietly turned off that very helpful feature. The only way I knew for sure that it wasn’t a glitch on my end was a post buried on X from Team YouTube.
So, email notifications are no more. There are other approaches. Most social network management tools include YouTube as a social network. You can log in to a social-network manager and look at what comments need to be handled, in the same way you can check if there are mentions of your company on X, Facebook, or Instagram.
Unfortunately, it quickly became apparent that this alternative scenario didn’t work for me. I’m not called to action by social media managers. I’m called to action by specific emails that land in folders I check every day. I live in email, so this makes the most sense for my work style.
Also: I built an app for work in 5 minutes with Tasklet – and watched my no-code dreams come true
In a normal world, it wouldn’t have mattered what works best for me. You takes what you’re given, and you loves it, right? But this is no normal world. This is ‘The Age of AI.’ And where there’s AI, anything is possible.
Before I go on, I want to apologize to anyone who might have left a comment on my YouTube channel. I normally make it a priority to respond, but since I wasn’t getting email notifications, I just assumed I wasn’t getting any comments. That’s now been fixed, so if you comment now, I’ll know. Read on to learn how.
So I AI’d me a solution
I MacGyvered a hack using Gemini. That’s the magical thing about AI. You can often use it to solve problems that previously would have been impractical or impossible.
Take this notification issue. YouTube does have API calls that make building something to solve this problem doable, but you’d have to write some fresh code. I’m a good enough programmer that I’m sure I could have done so.
In the pre-AI world, I could have, but I wouldn’t have. That’s because I don’t have the free time to justify a single-purpose use programming project that would take me three or four weekends to accomplish (at best).
Also: 10 ChatGPT Codex secrets I only learned after 60 hours of pair programming with it
But now, a few prompts to the AI, and about an hour to hook everything up, and I have a working solution.
As I walk you through how I did this task, please keep in mind how you can AI a solution out of almost anything. It’s not that I expect many of you will need email-based notifications for YouTube comments. But most of you reading this will want to create something to solve a problem that you alone want to solve.
AIs, such as Gemini and ChatGPT, can do that work if you collaborate carefully. I’m using the paid $20/mo Gemini Pro tier for this project because Gemini is getting better and better, but I could have used ChatGPT, Claude, Grok, or Copilot.
Is there an API for that?
I first looked at the user-level features that might have helped me generate notifications. I asked Gemini, “Does YouTube have an RSS feed for comments?”
I was told, “No. They actually used to provide comment RSS feeds, but that feature was permanently discontinued back in 2015. Today, the only native RSS feeds YouTube still supports are for new video uploads, leaving community interactions completely out of the loop.”
Also: 10 things I wish I knew before trusting Claude Code to build my iPhone app
Then I asked, “What other automated comment notification tools are there?” I was sent to a few commercial tools for managing YouTube comments, but outside of the email flow. You log in to a third-party dashboard and manage the comments. That’s not what I wanted.
But then I asked the magic question, “Is there any programming interface whatsoever to be able to monitor and extract YouTube comments?”
This question resulted in exactly what I wanted to hear: “Yes, there is an official programming interface provided by Google specifically for this purpose: the YouTube Data API v3.”
Gemini continued, “This is the exact same underlying architecture that powers the third-party automation tools and social media inboxes we discussed earlier. If you have some programming knowledge, you can bypass those subscriptions entirely and build your own custom extraction and monitoring system.”
Also: How to switch from ChatGPT to Claude: Transferring memories is easy
Then it asked the best leading question, ever: “Would you prefer I write a basic example script in Python that demonstrates how to extract your channel’s comments?”
Why, yes, I would.
The spec in one sentence
Here’s what I told Gemini. I said, “What would a Python script that does this look like? All it would have to do is poll every hour to see if there are new comments and then send an email with links to the comments.”
Gemini then wrote the script for me. The result requires inserting a free YouTube Data API v3 key from the Google Cloud Console and an app password for my YouTube account, but these are easy to get and use.
Gemini suggested, “You can run this script directly on your local computer, or let it run in the background on a cheap cloud server (like a Raspberry Pi or an AWS EC2 instance) so it monitors your channel 24/7.”
This output triggered an aha moment that I confess I hadn’t considered. I told Gemini, “I have a Linux box running a couple of Docker instances. Can I set up this Python script in the background running in Docker?”
Also: This new Claude Code Review tool uses AI agents to check your pull requests for bugs
Even though I’ve been coding in Python for some time now, and it’s both straightforward and powerful, I hadn’t thought of creating a Docker instance to run Python scripts automatically.
Just as a quick background, Docker is a container service. Basically, it’s like a virtual machine (VM), but without the machine. When you set up a VM, you’re configuring an entire computer emulator and an entire operating system installation. So, if you’re running a bunch of VMs, you have a lot of overhead (and potentially OS licenses) to deal with.
A container service like Docker adds the application-specific layer on top of any other operating system. You can run the same Docker container on a Linux, Mac, or Windows computer, and it will generally be happy. I’ve been using Docker for article archiving with Karakeep on an old 2012 Intel Mac mini. This setup has had no performance issues whatsoever.
Before I asked Gemini for configuration guidelines, I realized I wanted one additional feature in my code. I wanted my Python monitoring script to check if the YouTube API I’m using still exists and let me know if it no longer works.
Also: This AI tool turned my messy browser tabs into something actually manageable
I prompted Gemini, “It’s always possible that YouTube will deprecate the API code I’m using in this Python script. Extend the code to include a test for the API’s existence (possibly through a timeout test) that notifies me via email if the API is no longer available. Tell it to try a few times over the course of a couple of days before sending me that notification.”
It replied, “To do this, we are going to change how the script handles network errors. Instead of quietly ignoring them, the script will now keep a running tally of consecutive failures. Because the script checks every hour, 48 consecutive failures equals exactly two days of the API being unreachable.”
And then it gave me new code.
Hooking it all up
Preparing the code involved filling in the following lines:
YOUTUBE_API_KEY = "YOUR_YOUTUBE_API_KEY" CHANNEL_ID = "YOUR_CHANNEL_ID" EMAIL_PASSWORD = "YOUR_GMAIL_APP_PASSWORD"
For the first element, I went to the Google Cloud Console and created a new project. I selected APIs & Services, then Libraries, and then chose YouTube Data API v3. All that was needed was to hit the enable button. There’s a Create API key button. Pressing that button created a text key I brought back and pasted into my Python code.
Next up is the Channel ID. This is just a matter of going to the YouTube channel in question, then Settings under the profile, visiting Advanced Settings, and copying the ID.
Finally, I went to my Google account management page, searched for App Passwords, and created one for this new app.
Also: I turned casual selfies into professional headshots with Gemini
You don’t have to know how to do these steps. I merely told Gemini to walk me through the steps to get these three identifiers, and it did. The AI presented me with step-by-step instructions that I followed to get the job done.
Next, it was time to hook everything to my Docker instance, controlled by Portainer, a Docker management tool. I once again used Gemini to guide me through the steps. In addition to the Python code, Gemini had me create a very small requirements file and the Docker file, which defines the Docker instance.
Then Gemini gave me a few command-line instructions to type in and run. There were a few back-and-forth interactions where I told Gemini what was on the screen, grabbed a screenshot, and then followed Gemini’s instructions.
After a few minutes, I had my Docker instance up and running, and an active comment-monitoring system. If you want to check out what Gemini built, I posted the Python code to GitHub. Feel free to use it, but please don’t ask me for tech support. I built this quickly for my own use. If you need help, I’d recommend you reach out to Gemini.
So does it work?
Yes, it does. The app checks for comments once an hour. This morning, I had this message in my email:
That’s exactly what I wanted. You can certainly run a project like this with Claude or ChatGPT. This is the first time I’ve taken Gemini all the way from idea to a complete, working project. I have to say that I was impressed by using Gemini 3.1 and the Pro account.
The AI answered all my questions, interacted clearly with no hallucinations, and in less time than it took me to write this article, I built a custom Python-based solution for my YouTube channel.
I am seriously stoked.
You can follow my day-to-day project updates on social media. Be sure to subscribe to my weekly update newsletter, and follow me on Twitter/X at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at Instagram.com/DavidGewirtz, on Bluesky at @DavidGewirtz.com, and on YouTube at YouTube.com/DavidGewirtzTV.
