Making YouTube Videos Faster with AI Tools and Google Veo 3
Jun 2025 · AI & TechGoogle now has four AI tools that cover a big part of making a YouTube video: scripting, building presentations, making B-roll, and making thumbnails. They're all Google products, and most are free or cheap.
I walked through the full workflow in the video above. This post breaks it down by tool and covers what works, what doesn't, and where the credits go.
Step 1: Scripting with Google AI Studio
Google AI Studio lets you prompt the Gemini models directly instead of going through the consumer Gemini app. For scripts, this worked best for me. I pull YouTube transcripts from other creators who cover my topic, import them into AI Studio as context, and ask for a new script.
You aren't copying their videos. The transcripts show the model the depth, pacing, and tone you want, and it writes a new script. You still need to rewrite it in your own voice before you record, but you're editing a draft instead of starting from a blank page.
AI Studio is free up to fairly high usage limits. The more specific your prompt, the better the script.
Step 2: Building Presentations with Google Vids
Once you have a script, Google Vids can turn it into a presentation-style video with slides, stock media, and an AI voiceover, all from the script text. The generation is free, and the result is decent for explainer videos.
You paste your script into Vids, let it make an outline, then let it build the video. The AI selects stock imagery and formats each slide around the script sections. You end up with a structured video with narration that you can either use as-is or take into editing software to refine.
Vids won't replace a talking-head video you film yourself. It works well for explainer and educational videos where people come for the information.
Step 3: B-Roll with Google Veo 3 (via Flow)
Veo 3 is Google's video generation model. You access it through Google Flow (specifically the Scenemaker feature), and you generate video clips from text prompts. It works like an image generator for video. You describe a scene and get footage.
The quality is clearly better than earlier text-to-video tools. The clips still look like AI (a bit too smooth and a bit uncanny), but they're good enough for B-roll in a YouTube video: environmental shots, product illustrations, and abstract visuals.
A few things to know before you start:
- Veo 3 generations cost credits. Free tier access is limited; plan on some spend if you're generating more than a handful of clips.
- The "Jump to" feature in Flow doesn't keep audio across segments. If you want music or narration under AI B-roll, you're adding that in your editor after export.
- Be specific in your prompts. "A busy coffee shop in the morning" produces generic results. "Wide shot of an empty office at dawn, warm light through floor-to-ceiling windows, slight camera drift" produces something usable.
Bonus: Thumbnails with Google Whisk (Imagen 4)
Whisk is Google's image generation interface powered by Imagen 4. For thumbnails, you describe the image you want, upload a reference image if you have one, and generate. The results are good, close to Midjourney for photorealistic thumbnail ideas.
I use it more for drafting ideas than for the final thumbnail. You can make 10 options quickly, pick the best one, then use it as is or take it into Canva or Photoshop to add your text and branding.
Putting the Workflow Together
Here's the whole process from start to finish.
- Research topic, pull reference transcripts → AI Studio for script generation
- Edit and personalize the script → your voice, your take
- Feed script to Google Vids → presentation-style video with AI narration
- Generate B-roll clips in Flow/Veo 3 for any visual-heavy segments
- Draft thumbnail concepts in Whisk → finalize in Canva
- Record talking-head segments if needed → combine everything in your editor
Not every video needs every step. For a talking-head video, you might only use AI Studio for scripting and Whisk for the thumbnail. For faceless educational content, you might use all four and skip the camera entirely.
What I Think So Far
This process cuts my production time. The scripting help alone is worth the setup. I get from a blank page to a structured draft in 15 minutes instead of two hours, which is a big deal when you make videos alone.
The AI video generation is impressive for 2025 and will be better in six months. Right now, it's best suited for B-roll and supplemental footage rather than primary footage. Anyone paying attention can tell it's AI. If people watch your videos to see you, keep filming yourself and use these tools for drafts and B-roll.
For faceless channels and explainer videos, the tools are ready now. Making that kind of YouTube video got a lot easier.
If you're building an AI-assisted video workflow and want to compare notes, I talk about this in the AI Community Discord. You might also like the NotebookLM Video Overviews post, which covers another Google tool for AI-made video.
Related reading: NotebookLM Video Overviews: Google's AI Video Explainer · AI Video Editing for Short-Form Content
Tools worth your time
The tools and workflows I use, tested on my own projects.
No spam, unsubscribe anytime · Privacy
You're in. Check your inbox for a welcome note.