I teach platform-proof monetization: making money online without relying on a single algorithm or platform to pay you. Part of that is being present in many places without doing the work many times. So here’s the setup I built that turns one video into long-form LinkedIn posts, Reddit posts, Instagram images with captions, Twitter posts, and more, automatically.
It took a few hours to build and will take you a few too. Once it’s running, one upload feeds everything. This article walks through the exact flow, the tools, the prompt rules that keep it from sounding like a robot, the honest drawbacks, and a step-by-step path to build your own version.
What You’ll Walk Out With
- How one TikTok becomes posts on five platforms, hands-off
- The transcribe, summarize, generate, post loop
- The prompt rules that keep it sounding like you, not a robot
- The honest drawbacks (what breaks, what it costs, where it gets stuck)
- A 5-step path to build your own simpler version this week
- A free 2-minute quiz at finder.platformproof.com to find what to build content around
Why Platform-Proof Repurposing Matters
If you only post on one platform, you’re one algorithm change away from a quiet account. I’ve watched creators with 100k followers on a single app lose 80% of their reach overnight because the platform shifted what it pushes. Spreading content across five platforms is not about being everywhere for the sake of being everywhere. It’s insurance. When one app slows down, the others keep pulling people back to the thing you actually own (an email list, a digital product, a website).
The catch: doing it manually is a full-time job. Writing a LinkedIn post, rewriting it for Twitter’s character limit, designing an Instagram image, drafting a Reddit post that doesn’t read like an ad, then pasting links into Pinterest takes an hour or two per video. Multiply that by even three posts a week and the math stops working. That’s the gap automation fills.
The Core Loop
I use AI agents inside n8n, an automation tool. The flow:
- I upload a TikTok as normal. A repurposing tool spreads the video to YouTube, Instagram, LinkedIn, Twitter, and Pinterest, and saves a copy to Google Drive.
- A workflow watches that Drive folder, and when a new file lands, it sends it to Amazon Web Services to transcribe. (I prefer AWS over Whisper because it accepts both video and audio and has a higher size limit.)
- The transcript and a summary get saved to Air Table.
- Separate per-platform workflows check Air Table about every hour for a new entry marked “not posted.” When one’s found, an AI agent writes that platform’s post, posts it, and flips the record to “yes.”
I keep each platform in its own workflow. It’s called modularization, and it’s the cleanest way to get the most out of AI agents. If LinkedIn’s API breaks, only the LinkedIn workflow fails. Everything else keeps running. The alternative (one giant workflow that does all five) is a nightmare to debug, and one tiny error kills the whole thing.
The Per-Platform Specifics
LinkedIn: long-form post, 1,200-1,500 characters, three paragraphs, hook line first, story-first not tip-list-first. The prompt tells the AI to start with a contrarian statement so the post earns a stop while scrolling.
Twitter / X: 280 characters max, hard cap. If the agent’s output goes over, the workflow loops back and rewrites until it fits. I learned this the hard way: without the check, half the posts got rejected by the API for being too long.
Reddit: the trickiest. Reddit hates obvious self-promotion. The prompt tells the AI to write the post as a “I tried this, here’s what happened” story, no links unless the subreddit allows it, no emojis, plain language. Pick subreddits where you’ve already participated, otherwise auto-moderators flag you.
Instagram: image plus caption. I use Banner Bear (more on that below). The caption is 150-250 words and ends with a question, because comments lift the post higher in the algorithm than likes do.
Pinterest: the image flows in from Banner Bear too, the link points to a blog post on alstongodbolt.com, and the description is keyword-rich. Pinterest is a search engine, not a social feed, so the keywords matter more than the wording.
Not sure what content you’d be repurposing?
The free 2-minute quiz at finder.platformproof.com helps you find your topic, based on the skills you already have. Same email unlocks every other video’s worksheet.
The Prompt Rules That Keep It Human
Be very specific with the AI agent’s prompt. Mine:
- Uses my tone of voice. The easy way to capture yours: feed a few of your text or audio samples into the AI and let it describe your voice, so the output sounds like you, not a robot.
- Says plainly: no jargon, simple language.
- Writes at about a sixth-grade level. That’s roughly the average US reading level, and many readers worldwide read English below that, so simple includes more people.
- No asterisks for emphasis (AI loves them).
- Hard character limits per platform. For Twitter, I check the character count, and if it’s over 280, the workflow loops back and rewrites until it fits.
- A list of banned AI tells: “ecosystem,” “navigate,” “in the realm of,” “delve,” “tapestry.” If the AI uses any, the workflow rejects the output and asks for a rewrite.
Images, Automatically
For Instagram, I use a tool called Banner Bear. I upload a background image I made in Photoshop with a text box, and the AI agent’s output drops into that box to create a scroll-stopping image with a search-optimized caption. Instagram’s a little harder because it runs through the Facebook API, but you set it up once and just keep it from breaking.
If Photoshop is overkill for you, Canva works the same way. Make a template with a text placeholder, then connect Canva’s API (or use Bannerbear, which is friendlier for automation). The cost difference: Canva Pro is $13 a month, Bannerbear starts at $49 a month for 1,000 images. If you’re posting daily across two platforms, Bannerbear pays for itself in saved time.
Honest Drawbacks (What Breaks, What It Costs)
It breaks more often than you think. APIs change, free tiers fill up, n8n updates push a node into a new format. Once a month I open the dashboard and find one workflow stuck. Budget 30 minutes a week to babysit it.
Cost adds up. n8n (self-hosted): free. AWS Transcribe: a few dollars a month for my volume. Air Table: free tier works at first, $10-20 a month once you grow. Banner Bear: $49/month minimum. OpenAI or Claude API: $10-30 a month depending on volume. Total: roughly $80-110 a month for a setup that replaces two to three hours a day of manual posting. Worth it once you’re posting regularly. Not worth it if you’re posting once a week.
AI-written posts read flatter than your best human-written ones. Even with the best prompt, the auto-generated version is a B-minus version of what you’d write at your best. The math still works because the AI version exists and a perfect version you never wrote does not, but don’t expect viral hits from automated posts. Use automation for volume and presence, save your best writing for the cornerstone pieces.
Reddit and LinkedIn will sandbox you if you over-post. Both platforms penalize accounts that look automated. Cap the AI posts at two to three a week on those, and mix in genuine human comments and replies. The other platforms (Twitter, Instagram, Pinterest) tolerate volume much better.
Why This Matters
One piece of content now becomes text, images, and video across platforms, and I meet people where they are. Someone scrolling LinkedIn on mute still stops for an image. And because the setup already exists, I can spin up a whole new niche (I’m now in a “can dogs eat this” pet niche) without spending more time. The larger and more consistent your presence, the higher your chances of making money, and none of it depends on one platform paying you.
It’ll feel intimidating and frustrating the first time, expect about five hours and some self-doubt. But once you’ve connected the dots once, it gets easier every time, and there’s plenty of documentation (and ChatGPT) to help when you get stuck.
How to Build Your Own Simpler Version This Week
1. Pick two platforms, not five. Start with LinkedIn and Twitter, or Instagram and Pinterest. Two is enough to prove the system works without overwhelming you.
2. Set up n8n (or Make.com if you prefer a friendlier UI). Self-hosted n8n is free forever. Make.com has a free tier (1,000 operations a month) that’s enough to start.
3. Build the transcription step first. Upload a video, get a transcript out, save it somewhere. Don’t move on until this works end-to-end.
4. Add one platform-posting workflow. Pick the easier one first (Twitter or LinkedIn). Get one post live, then duplicate the workflow for the second platform.
5. Iterate the prompt for a week. Your first 10 auto-posts will be 60% good. Tweak the prompt every time something reads off. By post 30 you’ll have a prompt that hits 90%.
Find Your Topic First
Take the free 2-minute quiz at finder.platformproof.com. You’ll walk out with one specific thing to build content around.
Frequently Asked Questions
Do I need to be a developer to build this automation?
No. n8n and Make.com are both no-code (or low-code). You drag nodes onto a canvas and connect them. You will hit moments where you need to read API documentation or paste a snippet of code, but ChatGPT or Claude will write the snippet for you if you describe what you want. The real skill is patience, not programming.
How much does this stack actually cost per month?
Roughly $80 to $110 a month for the full five-platform setup: AWS Transcribe (a few dollars), Air Table (free to $20), Banner Bear ($49), an AI API like OpenAI or Claude ($10-30), and n8n free if self-hosted. A simpler two-platform version can run on $20-30 a month using free tiers.
Why use n8n instead of Zapier or Make.com?
n8n is self-hostable (free forever) and built for AI agents, with native nodes for OpenAI and Claude. Zapier is friendlier but expensive at scale and weaker on AI. Make.com sits in the middle: friendlier UI than n8n, cheaper than Zapier, with good AI support. Pick n8n if you’re comfortable with a bit more setup, Make if you want it easier on day one.
Can I start with one platform and expand later?
Yes, and you should. Start with one (whichever you actually use). Get the full loop working (upload to one platform, auto-post to another), then duplicate the second workflow for a third platform. Trying to build all five on day one is how people quit before anything ships.
How do I keep the AI-written posts from sounding like AI?
Three things: feed the AI a tone-of-voice sample (5 to 10 of your real posts), ban the obvious AI tells in the prompt (“ecosystem,” “navigate,” “delve,” em-dashes, asterisks for emphasis), and run a sixth-grade reading-level constraint. The biggest single fix is the tone-of-voice sample, you can paste your last 10 LinkedIn posts and tell the AI to match.
Will platforms ban my account for automated posting?
The risk varies by platform. Twitter, Instagram, Pinterest, and LinkedIn tolerate API posting through official integrations. Reddit is the strictest, treat it like a hand-curated channel and post less often. The fastest way to get sandboxed anywhere is over-posting or posting links without engagement. Cap your auto-posts and mix in real human comments.
What’s the most common reason setups like this break?
API token expiration. LinkedIn, Instagram, and Twitter all require you to refresh tokens every 30 to 60 days. If you don’t have a calendar reminder, one quiet month later you wonder why nothing posted. Set a recurring 30-day reminder the day you build it. Second most common: hitting a free-tier limit (Air Table records, AI API monthly cap) without realizing.
What if my niche is small, is automation still worth it?
Yes, possibly more so. Small niches reward consistency over volume, and consistency is exactly what automation gives you. If you’re posting once a week manually, automation gets you to three times a week without more time. In a small niche, that’s the difference between “I’ll check back later” and “I see this person everywhere.”
Read Next
Want to see one piece of this in detail? Here’s the Facebook-group posting agent, step by step.
Read: I Built an AI Agent to Post in Facebook Groups
Sources
- n8n (automation), Make.com (alternative), Air Table (content database), Amazon Web Services Transcribe, Banner Bear (image generation), Canva Pro (cheaper image alternative)
- OpenAI and Anthropic (Claude) APIs for AI agents
- Free 2-minute Side Hustle Finder quiz: finder.platformproof.com
Helping 1 million working adults make their first $3,000 online with the skills they already have. Alston Godbolt, Platform Proof.