One journal entry in the morning becomes a reel script, carousel outline, caption, and email subject line by lunch. Here's the exact prompt chain, broken down step by step.
I publish 5 pieces of content a day across Instagram. Reels, carousels, captions — all of it. With no team, no content agency, and no VA scheduling posts. The entire pipeline runs through AI, and it starts with the same thing every morning: a raw journal entry.
The secret isn't some magical prompt. It's the system underneath — a set of custom skill files I've built inside Claude that know my brand voice, my script formats, my hook patterns, and my content pillars. When I feed in a raw thought, the system produces native-format outputs for each platform because the rules are already baked in.
The gap between idea and published needs to be as short as possible. If it sits in a drafts folder for more than a day, it's dead.
Five minutes in the CTRL app. Stream of consciousness. Could be a frustration with a tool, an insight from yesterday's build, a story from my corporate days, or a contrarian take that's been rattling around. I don't think about format — I just get the thought out.
I paste the raw capture into my becomingscott Claude project. The project has 11 custom skill files loaded: brand voice, script writer (yapping and storytelling formats), hook generator, carousel builder, caption writer, and more. Claude already knows who I am, who the audience is, and what the brand sounds like.
From one raw capture, I ask for the full repurposing spread: a yapping reel script (under 40 seconds), 5 hook variations, a carousel outline (slide titles), a short caption, and an email subject line. Claude produces all of them in one pass because the skill files handle the format-specific rules automatically.
I read each output once. Sharpen the hooks — that's where most of the editing happens. Cut anything that sounds too polished or too "AI-written." Make sure every script passes the spoken word test: read it out loud, and if any sentence sounds written rather than said, rewrite it. Then film or schedule.
The output from this single prompt covers about 60% of my daily content needs. The remaining 40% comes from whiteboard instructionals and carousels that need more intentional planning — but even those start from the same journal capture, just with a different prompt.
Without skill files, Claude produces generic content that sounds like every other AI-generated post on the internet. With skill files, it produces content that sounds like me — because the files contain my actual voice patterns, my hard rules for what to avoid (staccato fragment lists, generic motivational language, guru-speak), and my specific format requirements for each content type.
Building the skill files took about a full day. They've saved me hundreds of hours since. The investment is front-loaded, and the compound returns are significant. If you're producing content regularly with AI and it still sounds generic, this is probably the missing piece.
Perfecting before publishing. The raw energy is the point. If I spend 30 minutes polishing a script that was meant to be unscripted, I've defeated the purpose and burned the time the system was supposed to save.
Batching too far ahead. I tried batching a week of content at once. By Wednesday the content felt stale because it wasn't connected to what I was actually building that day. The pipeline works best when the input is fresh and the gap to publishing is measured in hours, not days.
Ignoring what performed. The pipeline includes a feedback loop — high-performing reels get repurposed into carousels, and the hooks that stopped the scroll get recycled into new formats. If something resonated, I use it again from a different angle instead of constantly trying to invent new ideas from scratch.
Book a call. I'll either do the work myself, advise you through it, or connect you with the right person in my network. Every problem gets a path forward.
Book a Call