This comprehensive n8n workflow automate the handling of various media types received via Telegram, storing metadata, and generating AI-based descriptions and responses. The process begins with a Telegram trigger that captures voice messages, videos, images, documents, and other files. Files are then classified, processed, and stored in a PostgreSQL database, which groups media by IDs for subsequent batch handling.

The workflow includes several key steps:

– Media download and classification, with MIME type adjustments for diverse file formats.

– Media grouping, captions, and description storage in PostgreSQL tables.

– Trigger-based processing that waits for new media groups, retrieves stored descriptions, and compiles comprehensive media summaries.

– Advanced AI analysis using Google PaLM language models for images, videos, audio, and document content.

– Intelligent message formatting with MarkdownV2 for Telegram, splitting long responses into multiple chunks for seamless delivery.

This setup is ideal for building interactive chatbots, automated media summarization tools, or support systems that require multi-modal AI understanding and dynamic, context-aware conversations.

Practical use cases include customer support with multimedia messages, knowledge extraction from shared files, or creating rich, automated chat experiences powered by AI insights. The workflow also features database setup for persistent storage and mechanisms to handle multiple media files efficiently and reliably.

Node Count	>20 Nodes
Nodes Used	@n8n/n8n-nodes-langchain.agent, @n8n/n8n-nodes-langchain.googleGemini, @n8n/n8n-nodes-langchain.lmChatGoogleGemini, @n8n/n8n-nodes-langchain.memoryPostgresChat, aggregate, code, extractFromFile, html, if, manualTrigger, merge, postgres, postgresTrigger, set, stickyNote, switch, telegram, telegramTrigger, wait