I am building an automated marketing system in Make.com with the following objectives:
- Pull 3–10+ products dynamically from Airtable based on “Summer” suitability.
- Send those product details to OpenAI (gpt-4o or text-davinci-003) to generate concise British English marketing descriptions.
- Receive a structured JSON output per product (not one blob).
- Map each product individually into a Placid template (Product Name, Price, Marketing Description, Image).
- Email the final multi-product PDF to the marketing team.
Tech Stack:
- Airtable (Product Name, Price, Raw Description, Image URL, Weather Suitability field)
- OpenAI (via Create a Completion)
- Placid (for PDF brochure generation)
- Make.com (as orchestration platform)
Current Problem:
- OpenAI returns the marketing descriptions but they are stuck together inside one text block.
- I need to split them per product cleanly and map each product field dynamically into the Placid PDF.
- I already aggregate the Airtable products using Array Aggregator and prepare JSON using Text Aggregator and Set Variables.
Request:
What is the cleanest, production-ready way to:
- Send multiple products to OpenAI.
- Get clean, separate structured JSON responses (one object per product).
- Parse and map them into Placid correctly?
I am looking for a minimal-maintenance, scalable solution that handles variable product counts without hard-coding product slots manually.