Daily AI research summary from arXiv with Gemini AI, Notion & Telegram alerts

What this template does

This workflow turns the arXiv AI feed into a daily research assistant. Every morning it fetches the latest Artificial Intelligence papers from arXiv.org, deduplicates them, stores one page per paper in Notion (with metadata + PDF link), generates a Deep Research Summary of each PDF using Gemini, and finally posts a short update to a Telegram channel with links to both the paper and the Notion summary.

Who is this for & what problem it solves

This template is designed for founders, researchers, builders and curious professionals who want to stay up-to-date with AI research without reading every paper in full. It solves the “information overload” problem: instead of manually checking arXiv and skimming PDFs, you get a curated daily feed with human-style explanations stored in your own Notion database.

Use Cases

• Daily AI research digest for solo founders or small teams.
• Private “AI research hub” in Notion for your company or lab.
• Telegram channel that shares the latest AI papers plus plain-English summaries.
• Personal learning pipeline: track, tag and revisit important papers over time.

How it works (workflow overview)

1. Scheduled trigger runs every day at 08:00.
2. HTTP Request pulls the latest AI papers from arXiv’s API.
3. Results are converted from XML to JSON and cleaned.
4. A time-window filter keeps only recent papers and removes duplicates.
5. For each paper, a Notion page is created (metadata + PDF URL).
6. Gemini reads the PDF and returns a structured, multi-chunk summary.
7. Each chunk is appended to the same Notion page as rich-text blocks.
8. A Telegram message is sent with title, short abstract, PDF link and Notion link.

Setup (step-by-step)

1. Create a Notion database and connect your Notion integration.
2. Map the properties in the Register to Notion Database node (title, arxiv_id, abstract, authors, categories/tags, published date, pdf URL).
3. Add your Gemini API key and model in the Analyze doc (Prompt Ultra-Pro) node.
4. Add your Telegram bot token and chat_id in the Send a text message node.
5. (Optional) Adjust the arXiv query in HTTP Request to focus on your preferred AI categories or keywords.
6. Enable the Scheduled Daily Trigger when you’re ready to run it in production.

Customization options

• Change the arXiv search query (keywords, categories, max_results).
• Modify the time window logic (e.g. 24h, 48h, or no filter).
• Adapt the Notion properties to your own schema (status, tags, priority, etc.).
• Switch the messaging channel (Telegram, Discord, Slack) using similar nodes.