One thing I learned while building a Human-in-the-Loop AI automation system:
Good workflows are not just about automation.
They’re about control, clarity, and collaboration.
For my Relay challenge project, I designed the system using 3 separate workflows instead of one massive automation.
Why?
Because modular workflows scale better.
Workflow 1:
AI Research Pipeline
→ Generates hooks, pain points, content angles, and strategy insights.
Workflow 2:
AI Content Writing Pipeline
→ Creates platform-ready LinkedIn content after human approval.
Workflow 3:
Final Approval Pipeline
→ Handles final review and publishing status.
All workflows are connected through a single Google Sheet using status-driven automation.
Example:
Pending → Awaiting Approval → Approved → Awaiting Final Approval → Published
This architecture made the system:
• easier to debug
• easier to scale
• cleaner for human reviewers
• more production-ready
• better aligned with enterprise automation design
