Generate Blog Posts

Create polished, AI-assisted content from the processed product data.

Step 3
Generation Rate Limit Buffer

Each blog generation includes a 10-second safety delay between posts to adhere strictly to API rate limits (15 RPM). Please keep this tab open while processing a batch.

AI Copywriting Pipeline & Quality Gates

The blog generation service uses gemini-3.1-flash-lite (via OpenAI-compatible API) to transform raw product data into structured, luxury-oriented editorial posts:

  • Visual & Specs Integration: The model analyzes product titles, prices, descriptions, and up to 8 image URLs—embedding image tags directly near relevant visual discussions with custom call-to-action back-links.
  • Quality Control Gate (≥ 300 Words): Responses are stripped of HTML tags and word-counted. If an output drops below 300 words, the pipeline logs a warning and automatically retries generation up to 2 times.
  • Fail-Safe Fallback: If API attempts fail or word counts stay low, the system seamlessly falls back to a clean, structured HTML template (title, overview, features list) to ensure no queue stalls.
  • Throttled Processing: Standard POST runs clamp batches at max 20 items with a 10-second inline sleep between iterations to stay comfortably within free-tier API limits.
Generation Completion 100% Completed
Total Eligible
511
Generated
511
Pending
0