A faceless YouTube channel does not need a face. It needs ideas, scripts and a way to turn them into videos without losing a month. Claude is very good at the first two. Here is the whole pipeline, step by step, with the prompts I would paste.

I build Ziztu, so weigh that. The video part below runs on our AI YouTube Video Maker, and I will tell you where it is strong and where it is not.

What "30 videos in 30 minutes" means here. It means 30 films queued in about 30 minutes of your own time, roughly one minute of attention per film: paste, press Generate, read the script, approve. The films are then made on our servers while you do something else. I have not clocked 30 in a row, and every script you edit takes longer. It is a budget, not a promise.

What is a faceless YouTube channel?

A faceless channel is one where nobody appears on camera. The video is a narrator over pictures, maps, animation, footage or diagrams. You are not the product. The format is.

The common kinds:

  • History and true-story explainers. A real event, a narrator, maps and drawn scenes.
  • Science and "how does it work" explainers.
  • True crime and mysteries.
  • Sleep and bedtime history. Long, calm and rewatched.
  • Finance and business explainers, which tend to pay more per view.
  • Lists, rankings and "what if" videos.

The numbers people point to

I want to be careful here, because this is the part of faceless-channel content that is most often exaggerated.

  • The big channels are real. The channels people cite as faceless examples include Bright Side, Kurzgesagt, The Infographics Show and WatchMojo. One summary puts Bright Side's estimated AdSense revenue at $80,000 to $250,000 a month, and credits that to a third-party tracker (GoFaceless, Oct 2026).
  • Those are estimates, and they disagree a lot. Other trackers put the same channel far lower. Treat any single revenue number for someone else's channel as a rough range, not a fact.
  • Most channels start at zero. One 2026 breakdown says most faceless channels make $0 until they are monetized, then roughly $50 to $500 a month at 1,000 to 10,000 subscribers, $500 to $3,000 at 10,000 to 50,000, and $15,000 to $80,000+ a month only at 500,000+ subscribers (Korpi, 2026). Those are ranges from one source, not guarantees.
  • Long videos pay more than Shorts. The same source puts long-form RPM at about $2 to $15 per 1,000 views and Shorts at about $0.02 to $0.10. That gap is why 3 to 60 minute films matter.

The honest summary: the ceiling is high, the median is low, and the difference is mostly whether the videos are good and whether you keep making them.

One more fact before you build anything. YouTube says repetitive, mass-produced videos are not eligible for monetization (its "inauthentic content" policy, renamed in July 2025). Using AI is allowed. Publishing 30 videos that could be swapped for each other is the risk. Everything below is built around making each one different.

What each tool does

  • Claude (in Claude Code, a terminal, or a normal chat) does the thinking: the niche, 30 ideas, and a story brief for each.
  • Ziztu's AI YouTube Video Maker does the production: it writes the script from your brief, draws the pictures, reads it in a narrator's voice, and gives you the title, thumbnails and description.

Step 1: Ask Claude for 30 ideas you can check

Open a folder for the channel and start Claude Code in it (or use a normal Claude chat). Paste this:

You are my YouTube research partner. My channel is a faceless "history of engineering" channel. Videos are 8 minutes long. Give me 30 video ideas as a table with these columns: 1. Working title 2. The one true event the video is about, with a date and a place 3. The hook: who, when and where in two sentences 4. The single question the video answers 5. One source I can check (a museum, a book, an encyclopedia page) Rules: only events you are certain about. Mark any you are not sure of with "?". Never put two ideas from the same era next to each other. Save the table to ideas.md. 

Here are five rows in the shape the table should come back in. I wrote these from facts on our own history-ideas page, so use them as a format check, and verify every date for your own ideas:

# Working title True event Hook idea
1 Why the Pantheon dome never fell Pantheon, Rome, finished around AD 125 under Hadrian A dome 43 meters wide with no steel inside
2 The gear machine from a shipwreck Antikythera mechanism, found in a wreck off Antikythera in 1901 A bronze computer from the 2nd century BC
3 The stone that unlocked hieroglyphs Rosetta Stone, found 1799, carved 196 BC Three scripts, one message
4 How a baker's shop burned London Great Fire of London, from Pudding Lane, 2 September 1666 A fire that burned for about four days
5 Digging a canal between two oceans Panama Canal, opened 15 August 1914 A ditch that moved the sea

Do not skip the checking. Claude is good at ideas and sometimes wrong about dates. The "source I can check" column is there so you can spend two minutes verifying instead of publishing a mistake.

Step 2: Custom mode, where Claude writes the story and Ziztu makes the film

This is the part that makes each video different from the last. Instead of typing one line into Ziztu, you give it a real brief written by Claude.

Pick idea #1 and ask Claude:

For idea #1, write a story brief for an 8-minute film. - 4 chapters. Each chapter has a goal and ends on an open question. - The facts to include: numbers, dates, names, each with a source. - Plain chronological order. No flashbacks, no gimmicks. - Tone: warm storyteller. - Which chapters are about a place, so the film can show a map. - What to leave out. Under 400 words. Save it as brief-01.md. 

Now go to the AI YouTube Video Maker.

The AI YouTube Video Maker start screen with a box that takes a link or an idea, a Look picker, a Length picker and a Make my video button

The start screen. One box takes a link or an idea.

  1. Paste the brief into the box. Short briefs go straight in. If yours is long, paste the first lines and keep the rest of your notes for the Plan screen, where you can change any line.
  2. Set the length to 8 minutes. Any whole number from 3 to 60 works.
  3. Press Generate. You get a setup sheet for the look, the tone, the narrator and Basic or Pro, with the estimate in credits.
  4. Read the script. The tool stops once, at the finished script. You read it, change any line, and check the originality score. No picture is bought before you press "Make the film".

Pick the look so each channel feels like one place. There are 18, and one look carries every picture in the film:

The 18 looks: Map and cast, Old masters, Engraving, Cinematic real, Anime documentary, Paper cutout, Bedtime history, Cartoon comedy, Geometric Explainer, 3D Reconstruction, Archive Collage, Clay Stop-Motion, Brick World, Miniature World, Deep Space, Blueprint Cutaway, Dark Tech and Manga Ink

Looks. For an engineering history channel, Map and cast, Engraving or 3D Reconstruction fit well.

Then pick a narrator. There are 33, each with a short description and a sample you can play before you choose.

Step 3: The copy way, where you borrow a structure and tell your own story

The second way to start is to paste the link of a YouTube video that is already working. The tool does not copy it. It studies how it is built: the title pattern, the hook, the chapters and the pace. Then it writes your own version, about your topic, in your words.

Two safeguards make this honest:

  • An originality score shows how far your script is from the source, and marks the passages that sit too close. One click rewrites them.
  • Your own guide character. When a film copies a pasted video's drawing style, it gets its own simple guide figure, not the channel's mascot.

Five pictures drawn in the style of a pasted history video: a throne room, a guide next to a portrait, a winter river crossing, a second throne room and a relief map of Russia

A film drawn in the style of a pasted history video, with our own guide and a relief map.

A link only works when the video's captions can be read. If they cannot, paste the idea instead.

A good rhythm: use the copy way for the structure of your first video, then use Custom mode (Claude's brief) for the next ones, so your channel builds its own voice.

Step 4: Queue the other 29

Ask Claude for the rest in one go:

Write briefs for ideas 2 to 30 in the same format. One file each: brief-02.md, brief-03.md and so on. Vary the chapter structure so no two videos open the same way. Keep a facts-to-check list at the end of each file. 

Then work through them in Ziztu one at a time: paste, Generate, read the script, approve. This is where the "30 minutes" budget goes, at about a minute per film.

Three rules that keep it honest:

  1. Start with three, not thirty. Make three films and watch the price and the quality before you queue the rest. The price in credits is on the button before each film is made.
  2. Vary the looks and narrators across the batch, and do not upload 30 near-identical videos in one day. That is the "mass-produced" pattern YouTube warns about.
  3. Read every script. The tool flags facts it is unsure of, and you can ask for a web check, but you are the editor.

Step 5: The publish pack

When a film is done, you also get what you need to upload it: title options, three thumbnails, a description with chapter timestamps, tags, a pinned comment and an SRT captions file, in one zip. The topic appears on the thumbnail as a short label.

Download the MP4 and upload it yourself. If you chose the realistic look, use YouTube's label for altered or synthetic content.

A real example: a 5-minute documentary

Here is one film made with the tool, "9/11 and Its Roots: 32 Years in 5 Minutes". It runs 5 minutes in 4 chapters:

  1. A Tuesday Morning in Boston (0:00)
  2. The Road to Al-Qaeda (1:04)
  3. 102 Minutes (2:30)
  4. 3,520 Days (3:59)

The public watch page for the film, with the 5-minute video, its chapters and a Make your own button

The public watch page. Anyone with the link can open it, no account needed.

Watch the film

Three thumbnails made for the film: "9/11 1979 to 2011", "Boston Control" and "102 minutes"

The thumbnails that came with it.

It is a heavy subject, and it shows the structure that works for long films: a human-scale opening, a clear chain of cause, and numbers placed where they matter.

Honest limits

  • Claude is not a fact-checker. Use the sources it gives you.
  • Basic films are pictures with camera moves, not AI video clips. Real motion is Pro.
  • A link needs readable captions.
  • A film is not a channel. Titles, thumbnails, consistency and a point of view still decide whether anyone clicks.
  • Thirty is a stretch goal. Three good videos beat thirty forgettable ones.

Try it

You get 400 free credits and no card at getziztu.com/ai-youtube-video-maker. Ask Claude for ten ideas, pick one, paste its brief, and read the script it writes before you decide to make the film. If you make one, tell me what it was about.