Descriptions:
Creator Sharbel A. shares the complete architecture of “Sage,” a custom AI agent he built to solve a specific bottleneck in his Twitter content workflow: generating quality ideas consistently on days when back-to-back meetings leave no time to track what’s happening online. Crucially, Sage never posts autonomously — Sharbel is explicit that preserving his authentic voice means he writes every tweet himself — but the system handles research, trend detection, competitor analysis, and content ideation on a fully automated daily schedule.
Sage runs a multi-step morning routine: analyzing months of historical tweet performance to identify top and bottom 20% performers, scraping competitor posts, scanning Twitter for niche trends, and delivering reactive content opportunities via Telegram by 7:20 a.m. Three times daily it delivers five content ideas with angles and draft copy. Every suggestion receives an approve or reject response, and rejections require a stated reason — building a continuous feedback loop that Sharbel reports improved Sage’s idea approval rate from roughly 30% to 70-80% over eight weeks of use.
The quantified results ground the case study: a reCAPTCHA-themed post flagged by Sage reached 13 million impressions, a follow-up hit 2 million, and Twitter’s creator monetization program paid out $1,400 in a single month. The video also shows how Sage’s signals connect to a separate YouTube agent called Nova — routing audience demand signals cross-platform when a tweet generates unexpected interest in a topic. A detailed, reproducible example of multi-agent content operations in practice.
📺 Source: Sharbel A. · Published April 13, 2026
🏷️ Format: Workflow Case Study
