n8n: Turn Telegram Work Photos into Grounded LinkedIn Posts

Turn Telegram Work Photos into Grounded LinkedIn Posts—With Human Approval

This n8n workflow automatically converts a photo and caption from your Telegram bot into a professional, grounded English LinkedIn post using OpenRouter (vision + text), then sends you a draft in Telegram for approval before publishing to LinkedIn.

What this workflow does

• Triggers on Telegram messages when your bot receives a new message.
• Validates your input: checks that the message includes a photo and a caption of at least 20 characters. If incomplete, it prompts the sender in Telegram to provide what’s missing.
• Downloads the photo and converts it into a base64 data URL for an OpenRouter vision request.
• Extracts photo-grounded details via OpenRouter (Gemini vision): produces structured JSON including owner-reported facts, visible details, uncertainties, and possible post angles.
• Writes the LinkedIn draft via OpenRouter (GPT-4.1-mini): generates a professional English LinkedIn post in structured JSON, using the caption as the primary source and only photo details that directly support it (plus 3–5 hashtags).
• Sends a formatted preview to Telegram and waits for approval.
• Publishes to LinkedIn only after approval as a public personal image post; if rejected, the workflow stops without publishing.

Use cases

• Share daily work updates from a Telegram workflow inbox and publish them to LinkedIn with accurate, photo-backed wording.
• Turn field photos (site work, prototypes, events) into consistent LinkedIn posts while keeping a manual approval step.
• SaaS teams streamline “shareable wins” reporting without losing control of what gets posted.

Technical details

• Integrations: Telegram bot, LinkedIn (OAuth2), OpenRouter API (https://openrouter.ai/api/v1/chat/completions).
• OpenRouter steps: vision extraction with Gemini, then text generation with GPT-4.1-mini.
• n8n nodes mentioned: if, set, code, merge, Telegram, LinkedIn.
• Core data handling: photo → base64 data URL → vision request → structured JSON → grounded LinkedIn post draft.