← All Field Notes
· Sep 5, 2026

How I Started a Faceless YouTube Channel With 8 Prompts (Episode One Overnight)

Key Takeaways

I keep seeing the same post on Instagram. Eight prompts, comment a word, get the prompts, start a faceless page. Most people save it and never run it. I run On Point Tech Solutions out of Los Gatos, and my whole thing is putting AI to work instead of collecting it, so I loaded all eight into Claude on Fable 5.1 and told it to run them one by one for YouTube. This is what came out, including the part where it picked the wrong niche first.

Prompt One Pulls Real Demand, Not Guesses

The prompts came from a creator called Entrepreneur Dominate. His version is written for Instagram theme pages, so my only instruction was to adapt them to YouTube and follow them in order. What Claude did with prompt one is the reason the rest worked. Instead of listing niches it thought sounded good, it ran live YouTube searches, read the median views and checked how many of the top results one channel owns, then handed me one document.

That document is the difference between a faceless YouTube channel that finds an audience and one that talks to nobody. A huge top result sitting over a tiny median means people want the thing and nobody is serving it yet. A search owned by one channel means you would be fighting an incumbent on their own title. You cannot see either of those from a prompt that guesses.

Eight Prompts to a Channel page in Claude with the faceless YouTube niche research for prompt one

Round One Picked My Own Industry

The first pick was AI for trades and home services. Missed calls, screen recordings, bills, the whole receptionist angle. It is a good pick for what I do every day, and I am keeping the page it wrote, but I did not want a second On Point Tech channel. So I typed one more line and asked for something with no connection to my agency. Kids, technical explainers, news, anything.

Round two tested those three lanes against the same live demand data. Kids had the biggest numbers and every search was owned by a studio with an animation budget. Technical explainers were proven but the incumbents are 3D animation houses. News had a hole in it. Robotics news was owned by TV networks with zero creators, and nobody was covering the weekly creator show angle. It picked humanoid robots and named the channel The Humanoid Report.

What the Page Actually Wrote

Prompts two through eight build the channel it picked. Reading the page took me longer than running it. A few of the pieces:

Episode one of The Humanoid Report, the faceless YouTube channel Claude planned and cut overnight

One Line: Start This Project For Me

This is the part I actually wanted to test. I did not build anything by hand. I typed, "Hey, can you start this project for me? Let's create all the steps and execute," and went to bed. The next morning there was a six minute episode playing in my browser. It used my own cloned voice, pulled company clips of the robots, and opened with the numbers the episode is built around, Unitree shipping about 5,900 humanoids in the first half of the year and their cheapest one listed at $4,900 in stock. Those are the episode's claims, read off the sheet it built, not mine.

Was it perfect? No. I said on camera I would go through it one more time. For a one shot from a few prompts, it was better than most first episodes I have seen humans make.

Why This Matters for Bay Area and San Jose Businesses

Almost every business owner I talk to in San Jose and the South Bay says the same thing about content. They know they should be posting, they have no time, and they do not want to be on camera every day. A faceless YouTube channel is one answer, but the reason most of them die is the niche gets picked on vibes and the channel talks to nobody.

The thing to copy here is not the eight prompts. It is that the niche came from real search data, and the plan came with a system a part time VA can run. If you run a roofing company in Campbell or a dental office in Willow Glen, the same move works on your own industry. That first run, the one I threw out because it was my industry, is the one you would keep.

Practical Steps

  1. Get the eight prompts. They are free in the Academy, with my notes on adapting them to YouTube.
  2. Run prompt one against real search data, not from memory. If your AI cannot search, give it the numbers yourself.
  3. Read the niche pick out loud. If it is your own industry and you want your own industry, keep it. If not, say so in one line and run it again.
  4. Let prompts two through eight build the whole page before you edit anything.
  5. Type one line to start the project, then read every script and approve every upload yourself. Those are the two jobs the system should never take from you.

Watch the Build

We do not know until we run it. That was true of every automation I have ever built, and it is true of The Humanoid Report. If you want help building a content system that runs while you are on a job site, On Point Tech Solutions is a Go High Level consultant in Los Gatos serving San Jose and the Bay Area. You own everything we build. Done for you, or taught with an SOP so your team runs it. Start at optechsol.llc.

Want this working in your business?
Get my plan ← Back to all Field Notes