Build in Public

The problem every creator knows: "Just repurpose your content across platforms!" Sounds easy. In practice? Each YouTube video = 3-4 hours of: → Writing Twitter threads from scratch → Reformatting for LinkedIn → Coming up with platform-specific hooks It's a second full-time job.

I tried everything: ❌ Manually rewriting (slow, painful) ❌ ChatGPT + transcript paste (faster, but output = mid) ❌ Hiring a VA ($500-800/mo, inconsistent quality) None of it scaled. None of it was good enough.

So I built Replo. How it works: 1. Paste your YouTube channel URL 2. AI scrapes your videos + analyzes transcripts 3. Get 10 Twitter/X threads per week 4. Get 10 LinkedIn posts per week All formatted. All platform-native. All ready to post.

🧵 [SAMPLE OUTPUT — Generated by Replo] The most underrated skill in tech right now isn't coding. It's taste. Every company has access to the same AI tools. The differentiator is knowing WHAT to build and HOW to present it.

The playbook that's working right now: 1. Create ONE piece of deep, thoughtful long-form content 2. Extract 10-15 atomic ideas from it 3. Redistribute across platforms in native formats 4. Each piece stands alone but reinforces the whole This is exactly what Replo automates.

Frontend: Next.js 15 (App Router) + TypeScript + Tailwind CSS Why: → App Router for server components (fast initial load) → TypeScript catches bugs before users do → Tailwind = ship UI fast without fighting CSS files Verdict: No regrets. This is the meta for a reason.

AI Engine: OpenAI GPT-4o-mini Why 4o-mini instead of GPT-4o: → 10x cheaper → 3x faster → Quality is 90% as good for content generation → Custom system prompts per platform The trick: Platform-specific prompt chains, not one generic prompt.

Database: PostgreSQL (Neon serverless) Schema: → users, subscriptions, videos → generated_content, content_queue → youtube_channels, analytics 8 tables total. Started simple, stayed simple.