Multi-Agent AI System for Creative Production

I am building a multi-agent AI system to automate 80–90% of a paid creative production pipeline for Meta (Facebook/Instagram).

The system includes:

Research Agent (creative trend scraping + clustering)

Internal Performance Learning Agent (Meta Ads API + tracking integration)

Script Generation Agent (data-informed structured JSON output)

Static Creative Generation Agent

Video Generation & Editing Agent

Meta Marketing API auto-upload module

Performance Feedback Loop (closed-loop learning)

Slack-based control interface

Orchestration + memory layer with vector DB

This is not a prototype — it needs production-grade architecture, modular design, scalable cloud deployment, and performance monitoring.

I need help with:

System architecture design

Multi-agent orchestration strategy

Memory management to avoid hallucination/drift

API integrations (Meta Marketing API, tracking APIs)

Video generation pipeline design

Feedback loop learning architecture

Cloud infrastructure design

Scaling to 2,000 creatives/month

Performance optimization and monitoring

Goal timeline: 90 days (Phase 1)

I need guidance on feasibility, architecture decisions, and implementation roadmap.