Create a daily AI news digest from RSS with TypeSafe Jev, OpenAI, Telegram and Gmail

## Daily AI news digest from RSS with TypeSafe Jev and OpenAI Every morning this workflow reads your RSS feeds, lets TypeSafe Jev pick the stories that matter, merges duplicates across sources, has OpenAI write one-line summaries, then sends the digest to Telegram and Gmail and archives it in Google Sheets. Jev is a classifier that answers yes/no, score and choice questions for about $0.04 per million input tokens, so the cheap model filters 40 stories and the LLM only summarizes the best 10. A typical run costs well under one cent. ### How it works 1. Fetch: reads every feed, keeps stories from the last 24 hours, skips links seen in earlier runs. 2. Score: Jev rates topic relevance (0-1), importance (0-3) and category. 3. Merge: kept stories are compared in pairs; same-event stories become one entry with "also reported by". 4. Summarize: one OpenAI call returns a headline plus a sentence per story. 5. Deliver: Telegram message, HTML email and one Sheets row per story. Failed feeds, unscored stories and LLM errors are listed in the digest instead of stopping the run. ### Setup 1. Get an API key at typesafe.ai and add it as a Bearer Auth credential on both Jev nodes. 2. Connect OpenAI, Telegram, Gmail and Google Sheets credentials (delete any channel you do not need). 3. Fill the config node: topic, feeds, telegramChatId, emailTo, sheetUrl, language. 4. Activate the workflow. Link memory only works in active runs. ### Customization - Missing stories: lower relevantMin (0.5). Too noisy: raise it (0.7) or set importanceMin to 2. - Duplicates slip through: lower dedupMin (0.6). Unrelated stories merged: raise it (0.8).