n8n template · AI Inbox Pack

Get a daily email digest in Slack with n8n

Natural companion to urgency scoring, this workflow turns the non-urgent email queue into a daily digest: every morning at 8 am, an LLM reads the Supabase queue, ranks, summarizes and posts the result to Slack and Telegram.

By Clément Lacaille (Tech-Bharat) · updated 11 July 2026 · how we build and test

{
  "name": "FlowKit — Digest quotidien des emails",
  "nodes": [
    {"name": "Tous les jours à 8h", "type": "n8n-nodes-base.scheduleTrigger", "typeVersion": 1.2, "position": [-440,0]},
    {"name": "Emails en attente", "type": "n8n-nodes-base.supabase", "typeVersion": 1, "position": [-220,0]},
    {"name": "Agréger", "type": "n8n-nodes-base.aggregate", "typeVersion": 1, "position": [0,0]},
    {"name": "Rédiger le digest", "type": "@n8n/n8n-nodes-langchain.chainLlm", "typeVersion": 1.5, "position": [220,0]},
    {"name": "Modèle OpenAI", "type": "@n8n/n8n-nodes-langchain.lmChatOpenAi", "typeVersion": 1.2, "position": [220,200]},
    {"name": "Envoyer sur Slack", "type": "n8n-nodes-base.slack", "typeVersion": 2.2, "position": [460,-100]},
    // … 1 more nodes, connections, settings
  ]
}

The problem it solves

Batch-processing email is one of the best-documented productivity practices — provided the batch is readable. Opening thirty accumulated emails is still slow, even when none is urgent.

An LLM-written digest changes that: three lines for what needs action, one line to group the newsletters, and the morning starts with an overview instead of a dive into the inbox.

How the workflow works, node by node

Node names are in French, exactly as they appear in the imported workflow — the logic is language-agnostic.

1. 01

   Tous les jours à 8h

   Schedule (cron)

   Cron trigger (0 8 * * *). Change the expression for a noon digest, or twice daily by duplicating the workflow.

2. 02

   Emails en attente

   Supabase

   Fetches all emails_queue rows with pending status — the ones queued by the prioritization workflow since the last digest.

3. 03

   Agréger

   Aggregate

   Groups all items into one object: the LLM receives the full list in a single call, letting it group and rank globally.

4. 04

   Rédiger le digest

   LLM chain

   The prompt enforces a precise Slack format (priorities first, the rest in brief, 15 lines max) and a dedicated reply when the queue is empty.

5. 05

   Modèle OpenAI

   OpenAI model

   gpt-4o-mini is plenty for guided summarization; one call per day, negligible cost.

6. 06

   Envoyer sur Slack

   Slack

   Posts the digest to the channel of your choice, with native Slack formatting (*bold*, lists).

7. 07

   Envoyer sur Telegram

   Telegram

   Same digest to a Telegram chat — handy on the go. Delete this node if Slack is enough.

What you need to run it

• ✔n8n ≥ 1.60 with the LangChain nodes
• ✔An OpenAI API key
• ✔The Supabase emails_queue table being fed (by the pack's prioritization workflow, or your own inserts)
• ✔A Slack bot and/or a Telegram bot (BotFather) with its chat_id

Customization ideas

• →Mark emails as processed after sending by adding a Supabase 'update' node (status = sent) after the digest
• →Split the digest per team by filtering the queue on a department field
• →Send the digest by email instead of Slack by swapping the output node for a Send Email node