How to Make YouTube Videos With AI (Prompt to Published)

Seven browser tabs. That’s the honest starting point for most people who try to make YouTube videos with AI: one for the chatbot writing the script, one for the voice tool, one for the image generator, one for the editor’s timeline, one for YouTube Studio, and two more for the tutorials explaining why the audio is out of sync with the images. None of this is a rumor — it’s the default path, and it works. It’s just slower and leakier than the arithmetic suggests, and almost nobody accounts for the disclosure step until an upload gets flagged.

This is a real walkthrough of both routes: assembling the stack yourself, tool by tool, and using an integrated pipeline that does the same five steps in one workspace. Same output — an 8–20 minute narrated video — two different amounts of your Tuesday.

The five steps, whichever route you take

Every long-form AI-narrated video, regardless of tooling, passes through the same five stages:

1. Script — research the topic, write a narration script with scene breaks
2. Voice — turn the script into narrated audio
3. Visuals — generate or source images/video for each scene
4. Assembly — sync audio to visuals, add captions/transitions, export
5. Disclosure and upload — label the video correctly and publish

The DIY route runs these as four separate tools with manual handoffs. The integrated route runs them as one pipeline. Below, each step, side by side, with what it actually costs.

Grow with Alex, a YouTube automation channel with about 236K subscribers, walks the DIY stack start to finish in one screen-recorded video — script, voiceover, edit, thumbnail, upload — using free tools instead of paid subscriptions.

He generates the full script from a single ChatGPT prompt in what he calls “a matter of seconds,” then times the CapCut assembly step — dragging a voiceover and stock clips onto the timeline — at “five to ten minutes.” That’s faster than the estimates below because his route skips per-scene image generation entirely and drags in stock footage instead; it’s the trade-off the DIY route always offers: cut a step and you cut the time, at the cost of a video that looks like everyone else’s stock footage. Even in his fastest version of the DIY stack, the manual timeline work in step 4 is still the step he narrates most carefully.

Step 1: Script

Step 2: Voice

Step 3: Visuals

This is the batch-and-pick reality of the image step: a grid of variants comes back per prompt, and picking the closest match — or re-rolling — is the 2–4 minutes per image the DIY estimate above accounts for.

Step 4: Assembly

Step 5: Disclosure and upload

This step is identical either way, and it’s the one most tutorials skip.

YouTube requires creators to disclose “realistic” AI content — content a viewer could mistake for a real person, place, or event, including synthetic voices and AI-generated visuals presented as real. You disclose it in YouTube Studio’s upload flow, under Details, by selecting “Yes” under Altered content. Per YouTube’s official guidance, disclosing AI use does not limit a video’s reach or affect monetization eligibility — the label just appears in the description and, for sensitive topics, in the player.

The arithmetic, side by side

For one 15-minute video, first time through each tool:

The DIY total doesn’t include render time, re-exports after an edit gets missed, or the learning curve on four separate interfaces. It also doesn’t include what happens when you catch a factual error in minute 8 of a finished script: in the DIY stack, that’s a new script, new audio, and re-timed images. In a pipeline built around scene structure and per-scene regeneration, it’s one sentence and one image.

The dollar comparison isn’t as lopsided as the time comparison — you’re paying for tool access either way. Longform Studio runs $49/month on the Creator plan plus production costs billed in real dollars per video, not opaque credits, which lands in a similar range to stitching together four subscriptions once you’re producing regularly. The case for an integrated route isn’t that it’s cheaper in subscription terms. It’s that editing one sentence doesn’t force you to redo three other tools’ worth of work, and that source-backed research and per-scene edits are built into the workflow rather than something you reconstruct by hand every time the script changes.

If you’re deciding between the two, the real question isn’t “can I do this with free tools” — you can, and the DIY route is a legitimate way to learn what long-form AI video production actually involves. It’s whether you’re publishing one video to see if the format works, or ten a month where the compounding edit cost is the thing that actually determines how much you make. For background on the format itself, see how to make a long-form video on YouTube and, if faceless is the goal specifically, how to make faceless YouTube videos covers the parts specific to never appearing on camera. What counts as “long enough” in the first place is its own question — long-form vs. short-form content lays out how the 8–30 minute band affects retention and RPM. For a deeper look at end-to-end tooling, our long-form AI video generator guide covers the render and delivery side in more detail, and if a single all-in-one “generate video” button is what you’re actually asking about, the best faceless AI video generators roundup covers why most of those still need real cleanup afterward.