Hey n8n community! 
I’m looking for some feedback/advice on an Agent AI project I’m currently building.
Project Goal:
An AI Agent designed to generate highly personalized LinkedIn posts for store directors (e.g., Carrefour).
The content is based on:
- Upcoming Commercial Events (from a Google Sheet).
- The unique Persona/Editorial Style of each Director (critical for authenticity).
My Workflow Logic (Sub-Workflow):
The process runs through a sequential chain of “small agents”:
- AI Planificateur: Looks at the event calendar and decides the next post/director to target.
- AI Rédacteur: Writes the post, strictly following the director’s persona profile.
- AI Validation: Scores the post on compliance, style, and fidelity (scores are written to an ‘Evaluation’ sheet).
- AI Superviseur (Main WF): Reads the validation score and decides if an email (for manual approval) is sent to the director.
My Questions to the Community:
- Coherence: Does this architecture (
Supervisor WF calling a 
Sub-Workflow of sequential agents) seem coherent and robust for this type of task? - Simplification: If you had to simplify this setup, where would you start? Could the Plan, Write, and Validate steps be consolidated into a single, more powerful Agent without losing quality control?
Any feedback on the structure or the concept is highly appreciated! Does it work on? I don’t how to use the aggregate or function modules 
