Skip to content
Estudio de Canales

How to Make a Documentary for YouTube, Including with AI

· 8 min read · Estudio de Canales

Cover graphic for Estudio de Canales: horizontal bars on a black background, with the article's headline in Spanish

To work out how to make a documentary for YouTube, think of it as an investigation told as a story: a starting question, material that answers it (documents, testimonies, places, archive) and a narrative that moves towards a conclusion. AI can help you organise questions, write drafts, illustrate and narrate, but it cannot be the source of what you tell, and if it realistically recreates events that did not happen that way, YouTube asks you to disclose it.

What sets a documentary apart from other videos

A documentary is distinguished by its method more than by its length: it starts from an open question, relies on material that can be checked and reaches a conclusion the viewer can follow step by step. An explainer video answers something already known; a documentary investigates something that has to be reconstructed.

You can tell by three things:

  • A thesis or a question: “what happened to the factory that gave work to half the town”, not “industry in general”.
  • Documentary material: documents, press, photographs, blueprints, interviews or footage of the places. It is what turns the story into more than an opinion.
  • A narrative arc: set-up, development with twists, and a close that answers the question, or explains why there is no answer.

If what you want is to explain a concept clearly, the explainer video format may suit you better.

Before producing: question, research and permissions

Pre-production is half the documentary. Before writing a line of the script:

  1. Formulate the question in one sentence and decide what stays out.
  2. Make a list of what you need to find out and where each thing might come from: archives, newspaper libraries, academic studies, people who lived it or know it.
  3. Go to primary sources whenever you can, and note each fact with its origin.
  4. If you interview someone, explain what it is for and get their permission in writing to use their image and voice. How to word it depends on the case.
  5. Check the rights to every piece of material that is not yours: photographs, archive recordings, maps. Being on the internet does not mean it can be used.

With other people's material there is also a YouTube rule worth knowing: among what cannot be monetised, its monetisation policies cite content downloaded or copied from another online source that has not been substantially modified. A documentary is not a montage of other people's videos: archive material accompanies your narration, it does not replace it. If your documentary is on a historical subject, you will find guidelines for researching and for handling war or violence with context in our article on YouTube history channels.

The structure: from question to conclusion

A simple structure that works on YouTube has five parts:

  1. Opening: the question or the image that prompts it, in the first few seconds.
  2. Context: the minimum needed to understand what is at stake.
  3. Investigation: the clues, in the order that lets viewers understand the case, with a document or a testimony that changes the view halfway through.
  4. Conclusion: what can be stated from the sources and what remains open.
  5. Close: what remains today, what changed or why it matters.

A documentary is usually long, so divide it into chapters with clear titles. On YouTube, to add them by hand, the first timestamp in the description list has to start at 00:00, you need at least three in ascending order and no chapter can last less than 10 seconds, according to its help page on chapters in videos. Bear in mind the length too: by default, YouTube lets you upload videos of up to 15 minutes, and verified accounts can upload longer ones.

To turn the script into scenes, with their narration, image and duration, our guide to a storyboard for a YouTube video will help.

Documentary with AI: what it can and cannot do

Making a documentary with AI is possible if the AI works for your research and not in its place. It can help you organise the questions, suggest a structure, write script drafts, generate illustrations for what has no image and provide a narration voice. What it cannot do is give you the facts: an AI model can produce results that look plausible and are wrong, and even links to web pages that never existed.

A documentary made with AI also has two fronts on YouTube:

  • The AI disclosure. YouTube asks you to disclose the use of AI-generated content if it appears that a real person does or says something they did not do or say, if a recording of a real event or place has been altered, and if realistic scenes that did not actually happen have been generated. Among its examples that require it, it cites the realistic depiction of a tornado or other weather phenomena approaching a real city when this is not true. If the content is not realistic, there is no need to disclose.
  • Monetisation. YouTube's policies accept content that uses creative tools to help deliver a unique, well-researched or creative narrative, such as using AI to edit scripts. By contrast, content without a clear thread is not monetised, such as videos that combine unrelated or incoherent AI clips to surprise or shock.

The practical conclusion: in a documentary with AI, the research and the sources are yours, recreations are recognised as such and every image serves the story.

Image, voice and music

A YouTube documentary without your own filming is built from licensed archive material, maps, captions, illustrations and generated images. Still images come to life with gentle camera movements over the photo, like the Ken Burns effect.

Voice is the backbone of a narrated documentary: an unhurried pace, short sentences and the same voice from start to finish. Music accompanies the narration, it does not compete with it, and it must have a licence that allows its use; the options are in copyright-free music for YouTube.

What the Estudio de Canales does and does not do with a documentary

The Estudio de Canales is designed for videos narrated over images, and a documentary of that kind goes through its whole process:

  • Research guide. The AI prepares what needs to be found out, the key questions, the facts that will have to be checked against sources and the possible angles, without stating facts as true and without links. You can also write the research yourself with its sources.
  • From brief to script, with versions. The video goes through research, brief, outline, script, verification and approval. The AI only extracts the checkable claims from the script; all of them start without a source and become verified when a person gives them a source.
  • Long running time. The channel bible sets the target length in minutes, the storyboard allows up to 400 scenes per video, and the chapters come from the script sections and are validated against YouTube's rules.
  • Storyboard by scenes. The narration of the scenes comes, word for word, from the approved script, and each scene has its visual and a movement: Ken Burns, pan, zoom or static.
  • Voice and images. AI voice or your own, and images generated with AI or your own assets with their licence and provenance noted.
  • Cloud editing. A preview at 1280×720 and, once you approve it, the master at 1920×1080 and 30 frames per second, in landscape.

What it does not do: it does not generate video clips with AI (the storyboard can suggest a clip for a scene, but today it is not generated: that scene is resolved with an image or with a licensed video of your own), it does not add music to the edit, it does not record interviews or search archives for material, and it does not supply or check sources by itself. Nor does it publish: you download the package and upload it yourself in YouTube Studio.

Sources consulted

Checked on 11 October 2026. We link to the original source so that you can consult the current text.

Frequently asked questions

What do I need to make a documentary for YouTube?

A specific question, material that answers it (documents, press, photographs, interviews or footage of the places), permission for whatever is not yours and a structure that leads from the set-up to the conclusion. With good narration and licensed material, you do not need to film anything.

Can you make a documentary with AI?

Yes, if the AI supports your research and does not replace it: organising questions, script drafts, illustrations and voice. The facts have to come from sources, because a model can produce results that look plausible and are wrong.

Do you have to disclose on YouTube that a documentary is made with AI?

You have to disclose if it appears that a real person does or says something they did not do or say, if a recording of a real event or place has been altered, or if realistic scenes that did not actually happen have been generated. If the content is not realistic, there is no need.

Does the Estudio de Canales generate the documentary's videos with AI?

It does not generate video clips with AI. Each scene has its narration and a visual, such as a generated image or a licensed asset of your own, with a camera movement; the editing is done in the cloud.

Related articles