Build a faceless content pipeline with Claude Code. Pull videos, write captions, and watermark them, all in plain English.
This is the brain that runs everything. It runs inside Visual Studio Code and takes plain English commands. You tell it what to do in words. It downloads videos, watches them, and writes captions. No coding needed.
Anyone who wants AI to do real work, not just chat.
Pick a source platform and grab posts with an API. In the demo the source was X. Claude Code calls the API, finds the videos, and saves them to a local folder. In the demo this pulled 8 videos in one run. You point it at the source.
Use the [SOURCE] API to fetch the latest [NUMBER] posts. Download each video to a local folder. List the files when you're done.
Claude opens each clip and writes down what happens in it. It saves one description file per video in the root folder. It also flags anything explicit so you stay safe. In the demo all 8 videos came back clean. This gives you raw context for captions later.
Watch each video in [FOLDER NAME]. Create a description file for each one in the root folder. Flag any video with explicit content.
Use a predictive brain model as your caption guide. In the demo this was Meta's Tribe V2 model. It studies how the human brain processes content. Claude researches the model, then writes an Instagram caption for every video. It adds hashtags too.
Research how the [BRAIN MODEL] predictive model works. Use those principles to write an engaging Instagram caption for each video in [FOLDER NAME]. Add relevant hashtags.
Creators who want captions tuned for reach, not guesswork.
Brand every clip with a small logo in one corner. First take a screenshot of one video as a layout reference. Paste it into Claude so it knows where the logo goes. Test on one video, fix the spot, then apply it across the rest.
Add the logo [LOGO FILE] to the top corner of each video in [FOLDER NAME]. Use the screenshot I pasted as a placement reference. Do one video first so I can check it.
Keep your captions and files in one simple table. Use a notes or productivity tool you like, such as Notion or Obsidian. Build a content table that holds the video, the caption, and other details. It keeps you organized once you scale.
Anyone posting more than a few videos a week.
Not sure where to start? Use them in this order.
Start with Claude Code in VSCode. Open it and run one small command first. Then do Step 1 and pull a few videos before you try the rest.
Jump to Step 3 and Step 4. Wire the brain model captions and the logo watermark into your current flow. Then add the content table to stay organized.
🚀 Want the step by step build, plus the extra tools for auto-posting and analytics? I share it all inside Build with AI. Come build with us: skool.com/ai-for-everyone