Skip to content
Source posts workflow: scraping and ranking reels

Instagram RAG system

ShippedAutomation

Instagram content ideas grounded in the client's own knowledge base, approved from Telegram.

The problem

Generic AI ideas ignored everything the client actually knew. They needed ideas built from their own material, with a human yes before anything was written.

What we built

  • Source post scraper that pulls high-performing reels and ranks them
  • Breakdown and analyzer that reads each video for hook, narrative and CTA
  • RAG refinement: a set of agents that rework the idea against the client's knowledge base
  • Approval in Telegram, then a full script written on yes
  • Knowledge-base manager that keeps the vector store current

Inside the build.

Analyzer workflow reading each video
Analyzer workflow reading each video
RAG refinement agents against the knowledge base
RAG refinement agents against the knowledge base
Telegram approval and script generation
Telegram approval and script generation
Vector knowledge-base manager
Vector knowledge-base manager
“I told him what my vision was and he made it into a reality within 24 hours. His speed and explanations are insane.”
Rohit Iyer, Founder, Clients Forge

Need something like this built?

Book a call (opens the booking page in a new tab)
Next: Ramada by Wyndham, Ghaziabad Vasundhara