How I'm Building an AI-Native YouTube Channel
Most people never start a channel because the operations around content feel heavier than the content itself: no ideas, no research process, slow editing, messy analytics. In this walkthrough I show you the actual systems running this AI-native YouTube channel, from turning real conversations into content items to tracking every video's performance, with AI removing the drag while my own experience stays at the center.
1. The four phases (0:00)
I break the channel down into phases: first, how to get ideas to create content. Second, how to build the content with AI, meaning the structure and the video editing. Then how to store and keep track of the analytics on how your content is performing. Each phase gets its own system.
2. Ideation from calls and conversations (0:29)
Ideation is where most people stall; they never get started because they do not know what to talk about. To fix this, I built a system that takes in my calls, with friends, with customers, discovery calls, sales calls, and breaks them down into content items.
The result is a Notion page, pretty much an Excel sheet if you want, holding everything I have ever said about important AI subjects or my personal experience. From one single call you can get tens of ideas on what to post, each broken down by type: is it personal experience, is it AI, is it both? Those items then become LinkedIn posts or YouTube videos. The great thing is that it all comes from your precise experience, so it is never going to be wrong. You are the one narrating the story. And this works for any niche out there.
3. Research trends from competitor channels (1:39)
Sometimes you are just getting started and genuinely do not know what to talk about. That is where system two comes in: instead of mining your experience, you research what is trending. I keep a YouTube inspiration base in Notion with the channels I follow around AI and automation, channels with a ton of views and a ton of videos that can spark ideas.
The system grabs all of their YouTube videos, tells me what is performing well, what is missing from my own channel, and how I can take some transcripts and create that content myself.
4. Track competitor videos and comments (2:43)
Beyond the channel list, you can also keep a huge base with literally all the videos. In my case that is thousands upon thousands, which may be overkill; just taking the videos from the last one to three months is already extremely powerful. The bonus is the comment sections: comments are usually questions, so they hand you even more content ideas that are directly linked to what people actually want.
5. Film and edit with AI (3:20)
Once research is done, you film yourself and edit the video. Editing on this channel is quite lightweight right now, but that is changing, because AI has become fantastic at editing video. I can literally say a few keywords during filming, which I will not say in the video itself, to mark where sections should be cut or where transitions should go. You film with the camera, hand everything to AI, and it does the editing, plus the upload, plus the thumbnail. That is how this channel operates right now. The thumbnail step now has its own walkthrough: build an AI thumbnail generator from your real photo.
6. Track YouTube analytics (4:12)
Filming is great, but you need to know how content performs and what your baseline is. I built a system that tracks all of my YouTube videos and their performance: views, impressions, which hooks are doing well, which thumbnails are doing well. Inside my content management system, which is what CMS stands for, I can watch the evolution of my YouTube subscribers over time, and the evolution of my LinkedIn followers as well, day by day.
7. Build around your own experience (5:04)
The goal of all of this is not to replace the creator. AI is the assistant while you stay at the core of what you publish: your experience, your face. I am not creating an AI avatar right now, because I do not think that is a good way of maintaining contact with human people, even though the technology is really there and you can create fantastic avatars. For now, it is just me showing you how to build cool things with AI over the coming months.
Pitfalls and tips
- Do not wait for ideas to appear. Mine your own calls first; one conversation yields tens of content items.
- Recent beats exhaustive. Tracking competitor videos from the last one to three months is enough; you do not need thousands of videos to spot trends.
- Read the comments. They are questions in disguise, and they tell you exactly what your audience wants next.
- Keep yourself in the content. Automate the operations around it, not the experience at the center of it.
Where to go next
The editing and upload side of this pipeline runs on agentic tools; see which ones in The Two Agentic AI Tools You Need to Learn. For turning your live business data into dashboards like my CMS, watch Build Live Artifacts in Claude Cowork.
