Daily industry research library in Notion

Build me an evolving research library in Notion that captures every meaningful new article in my industry. Run on a cron, every day at 08:00 in my local timezone.

At each run, use Ahrefs Firehose Stream with recent:24h against a rule that covers my industry keywords. I will supply the Lucene query when I configure the workflow, for example: (title:"vector database" OR added:"RAG" OR added:"retrieval augmented") AND (page_type:article OR page_type:guide) AND language:en. Treat that query as the rule and just stream the matches for the last 24 hours.

For each match, read the page title, URL, source domain, and a snippet of the added text from the Firehose event. Then decide what kind of article it is: thought-leadership essay, product launch, how-to, benchmark, or news. Drop pure SEO listicles and shallow roundups. If it is none of the keeper categories, skip it.

Before writing anything to Notion, call Notion Query a Database against my target database, filtered by the URL property, to check whether this article is already saved. If it is, skip it. This dedupes across days when the same article keeps appearing in the firehose.

For each surviving item, call Notion Create a Page in my target database. Set these properties: Title, URL, Source Domain, Theme (a short phrase you infer from the content, like "RAG evaluation" or "vector index pricing"), Article Type (one of: thought leadership, product launch, how-to, benchmark, news), and Published Date. In the page body, write a 2 to 3 sentence why-this-matters summary and quote the single most interesting line from the article.

At the end of the run, log how many items the firehose returned, how many were kept, how many were skipped as duplicates, and how many new pages were created in Notion. I want this workflow built as an agent because the quality filter, theme tagging, and why-this-matters summary all need real reasoning.

Inputs I will provide at setup time: my Lucene query for industry keywords, my Notion database ID, and the timezone for the 08:00 cron.