Pinterest Publishing Automation

An educational brand already publishes its posts automatically through the AI Content Factory. This workflow adds Pinterest as a further channel. It picks up posts that are live on the other platforms and have not reached Pinterest yet, so no new editorial work is needed.

How the workflow runs

• A schedule starts it on Tuesdays, Thursdays and Saturdays at 10:40, and a NocoDB query finds the oldest post that has no Pinterest status yet.
• The post is routed by category (words, word parts or idioms) and its content is loaded and normalised into one shape.
• Title, description and alt text come from templates in a config node. When the text exceeds Pinterest’s limits of 100, 800 and 500 characters, the origin section is shortened first, so the meaning and hashtags survive.
• The first slide is found in Seafile and passed to Pinterest through a keyed proxy, because Pinterest fetches images by URL and the Seafile library is private.
• The Pin is created through the Pinterest API v5, the post’s Pinterest status is updated, and Telegram gets a success message.

Built to run unattended

• Every step from building the text to creating the Pin has an error path: the post is marked as failed with the reason, and Telegram gets an alert.
• A post with a Pinterest status is never picked up again, so a repeated run cannot publish it twice, and a failed post waits for a person to look at it.
• Board, link, hashtags and text templates live in one config node, so nothing else changes when they do.

The integration is registered on Pinterest as Shinkarenko Content Publisher, a private app I operate. It connects to one account through Pinterest OAuth.