From recurring research to structured output.

I designed and built an n8n workflow that checked financial-news sources and selected finance-focused YouTube channels on a schedule. New material became structured records in Notion, then fed AI-assisted content branches.

The original prototype included Gemini-assisted copy, Placid visuals, Naver Blog preparation, and Gmail run notifications. It was built and run locally.

Original n8n workflow canvas showing research collection, AI processing, Notion storage, and channel-specific content branches

Original workflow artifact. Open the image for a closer look.

Why these tools

n8n made the integration steps visible. JavaScript handled intermediate data; Notion provided a structured destination. Gemini supported draft creation, and Placid handled visual templates.

Process

Monitor sources → Collect links and data → Structure records → Generate content → Prepare delivery and notifications.

Business relevance

The project explores how a recurring research task can become a reusable process. Its value is in connecting source collection and downstream production without rebuilding the same steps for each item.

What I learned

  • Update checks matter: the same source should not be processed repeatedly.
  • Structured intermediate data helps AI output work reliably with JavaScript and external APIs.
  • Separating research collection from channel-specific production makes the workflow easier to inspect.

Validation & current status

Previously operational as a prototype. The retained workflow image documents its structure, but the localhost instance was not reachable during this portfolio review. Credentials, API versions, and individual branches have not been retested.

No production uptime, audience growth, or time-saving claim is made here. A rerun would need duplicate-handling checks, output review, and confirmation that all external integrations still work.

The proposed weekly research journal uses a separate approval step before publication. The historical content prototype is not evidence that that new review process already runs.