AI Content Creator with Self-Evaluation Loop: Perplexity Research → AI Writing → Quality Gate → Loops Publishing

Problem:

A content operation needed not just automated content generation but a quality assurance layer built into the pipeline, AI-generated content that didn't meet the standard was being published anyway with no mechanism to catch and rewrite underperforming drafts before distribution.

Solution:

Built a dual-trigger n8n content pipeline (chat + schedule) that sets prompts and brand configurations, splits by brand, pulls live research via Perplexity API, aggregates data, and routes content through a two-stage AI evaluation system, the RVSCL AI Agent (CC) powered by RVSCL Writer generates the content draft, which is then passed to the UB Eval Model v1.3 powered by RVSCL Reasoning v1.3 with structured output parsing to score quality; if content passes the quality gate it is sent to Loops API for email distribution, and if it fails it is flagged as IsRewrite True and looped back for regeneration.

Result:

Delivered a self-correcting content pipeline that generates, evaluates, and only publishes content that meets quality standards, failed drafts are automatically flagged and rewritten without any human review, ensuring every piece sent via Loops meets the required quality bar.