UK Regulatory News Filter & Approve n8n Workflow (AI)

Turn UK regulatory news into an “approved package” your team can trust—automatically

This UK Regulatory News Filter & Approve n8n Workflow (AI) reads article links from Google Sheets (and can also pull the latest FCA RSS items), removes duplicates, then uses Hugging Face or OpenAI embeddings plus Firecrawl and TypeSafe JEV scoring to approve the most relevant stories. The workflow writes both an Approved Package and a full outcomes ledger back to Google Sheets for clean, auditable operations.

What this workflow does

• Runs manually or via the included webhook, loading configuration such as Google Sheets tabs, rule thresholds, and the chosen embedding provider.
• Optionally clears the “Approved Package” and “Articles” ledger tabs in Google Sheets to keep each run consistent.
• Pulls source articles from the Google Sheets “Articles” ledger and, if enabled, fetches the latest items from the FCA RSS feed, mapping them into the same article format.
• Deduplicates by URL by filtering out links already present in the Google Sheets ledger, then records new URLs as pending.
• Pre-filters noise using deterministic keyword and source weighting rules, and batches survivors for AI embedding.
• Embeds and clusters using Hugging Face Inference API or OpenAI Embeddings, grouping near-duplicates by cosine similarity and selecting a primary story per cluster.
• Scrapes and scores: for non-hard-win primary stories, it uses Firecrawl to fetch content and TypeSafe JEV to generate structured relevance/classification scores.
• Writes results back to Google Sheets, including the approved story package and an outcomes ledger.

Use cases

• Compliance and risk teams need a weekly/real-time shortlist of UK regulatory updates without manual triage.
• SaaS operators want an automated “source-to-decision” pipeline that documents what was approved and why.
• Automation engineers need an n8n workflow that combines rules, embeddings, scraping, and scoring into a repeatable process.

Technical details

• Google Sheets (read article links; write Approved Package and outcomes ledger)
• FCA RSS (optional live intake)
• Embeddings: Hugging Face Inference API or OpenAI Embeddings + cosine similarity clustering
• Scraping: Firecrawl
• Scoring: TypeSafe JEV
• n8n nodes/logic: if, set, code, no op, merge, filter