I'm planning to build a white-label AI social media publishing system for my agency clients, and I'd love some advice from people who've already built something similar.
The idea
I already build websites and run Google Ads/Meta Ads for local businesses. Most of my clients also post daily on Instagram, Facebook, TikTok, and sometimes LinkedIn. Right now, they manually upload the same photos/videos and captions to every platform, which takes a lot of time.
I want to build a simple web app where the client only needs to:
- Upload photos or videos
- Write a caption (or let AI improve/generate it)
- Select the platforms
- Click Publish
Everything else should happen automatically in the background using n8n.
Planned architecture
Frontend (Lovable / Next.js / Replit)
↓
Webhook to n8n
↓
OpenAI (caption enhancement, hashtags, CTA)
↓
Publish to selected social platforms
↓
Save logs to database
↓
Return success to the frontend
The client never sees n8n—they only interact with a clean dashboard.
My biggest questions
- What's the best way to let clients connect their social media accounts?
- Instagram
- Facebook
- TikTok
- LinkedIn
- X
- Google Business Profile
- How does the OAuth flow usually work for a SaaS like this?
- Where should I securely store access tokens and refresh tokens?
- Does n8n handle token refresh automatically, or do I need to build that logic myself?
- Would you recommend using n8n as the backend for production, or would you put another backend (Node.js, Supabase Edge Functions, etc.) between the frontend and n8n?
- For agencies serving multiple clients, what's the best architecture for keeping each client's credentials and posts isolated?
If you've built a similar social media publishing tool or SaaS, I'd really appreciate hearing how you structured it and any lessons you learned along the way.
Thanks in advance!