Case 03 / 06
PowNed
PowRank digs up the news so the PowNed editors don't have to. An n8n workflow pulls in articles and lets AI score them the PowNed way. A dashboard I vibe-coded in Lovable shows the ranked list.
- Client
- PowNed
- Role
- AI, automation & dashboard
- Tools
- n8n, Lovable, Webhooks, OpenAI GPT-4o-mini, Supabase, RSS
- Focus
- AI · Automation · n8n · Editorial Workflow · News Sourcing
[ A ]Challenge
Editors spent a lot of time manually searching for potential stories. The question: can an automated workflow do the searching, and judge the news the way PowNed would, so journalists can focus on what's worth investigating?
[ B ]Process
- 01
News sources
A button in the dashboard fires a webhook; n8n pulls in articles from RSS feeds, cleans them up and filters the stream.
- 02
AI filter
Batches of articles go through an LLM chain (OpenAI GPT-4o-mini) with a structured output parser, so every article comes back as clean, scored data.
- 03
Relevance
Each story is scored through the PowNed lens (controversy, topicality, humour and 'plopkap') and compared with earlier rounds so only new articles are stored.
- 04
Editorial selection
PowRank ranks the results. Editors filter by category and province, read the editorial tip, and approve or reject each story.
- 05
Journalist
The selected lead goes to a journalist to investigate.
[ C ]Technology
n8nLovableWebhooksOpenAI GPT-4o-miniSupabaseRSS
[ D ]Solution
PowRank: a dashboard built with Lovable that triggers an n8n workflow via a webhook. AI scores incoming news on PowNed-worthiness, and the editorial team gets a ranked, filterable shortlist instead of an endless feed.
