AirtableWhat they did
The poster is building an SEO content pipeline in stages and automated the first stage, long-tail keyword discovery. A seed keyword lives as a record in Airtable. A webhook call starts the n8n workflow, which looks up the record, marks it in progress, and runs three research paths against YepAPI: related keywords, keyword ideas and autocomplete suggestions. The autocomplete results go through a further keyword-data request. Everything is written back to Airtable as one record per keyword, linked to the seed, with search volume, competition, keyword difficulty and CPC where the API returns them, and the seed's status is updated at the end. The workflow is published on GitHub.
How it went
One execution on the seed phrase about self-hosting n8n produced 26 raw keyword items: 7 related, 6 ideas and 13 from autocomplete. The poster is clear these are raw counts with duplicates, and that de-duplication and grouping by search intent are the next jobs. The following stage, subtopic research for article planning, is not built yet. The thread drew two comments.
Worth knowing
There is no language-model step in this workflow; it is a deterministic chain of API calls and Airtable writes, which is why it is cheap and repeatable. Which metrics you get depends on what YepAPI returns for a keyword. Treat the Airtable records as a research queue to review, not a finished keyword list.
Meta
Higgsfield.ai