Telegram Frontline Worker Onboarding with DeepSeek & Notion (n8n)

Onboard frontline workers in Telegram—automatically verify ID documents with DeepSeek and track everything in Notion (n8n)

This n8n workflow turns Telegram into a complete employee onboarding channel: workers submit ID documents in chat, AI extracts and verifies key details with DeepSeek, and Notion becomes a live HR dashboard—no HR logins, no manual document checking.

What this workflow does

• Routes every Telegram message based on message type: /start, document/photo upload, the word AGREE, or other content.
• /start onboarding: searches a Notion database using the sender’s Telegram chat ID. It either creates a new onboarding record and sends a document checklist, or sends a “welcome back” message with the current onboarding status.
• Captures uploaded documents: downloads the file from Telegram and branches by file type (PDF vs image).
• Extracts document content: extracts text from PDFs, or base64-encodes images, then sends the content to DeepSeek (chat/completions) to return structured JSON (document details + a readability verdict).
• Updates Notion + attaches originals: writes the extracted summary to the worker’s Notion record, sets status to Docs Verified or Docs Submitted, and uploads/attaches the original file to the Notion page using Notion’s file upload API.
• Policy acknowledgment in chat: replies with a policy summary and prompts the worker to respond with AGREE.
• Continues onboarding flow: when the worker replies AGREE, it updates status to Policy Acknowledged and sends the worker shift schedule (and supports daily reminders for anyone stuck mid-process).

Use cases

• Hourly and frontline onboarding where staff only have mobile access via Telegram
• Reducing HR workload by automating document intake, AI review, and status tracking in Notion
• Ensuring compliance by forcing policy acknowledgment before proceeding

Technical details

• Integrations: Telegram (message trigger + file download/upload), Notion, DeepSeek
• n8n nodes/tools: if, set, code, filter, notion, switch
• AI output handling: sends extracted content to DeepSeek chat/completions and processes returned structured JSON to update Notion workflow status