Telegram Knowledge Base Bot (n8n) with OpenRouter + Supabase

Turn your Telegram into a controlled, knowledge-grounded Q&A bot—powered by n8n, OpenRouter, and a Supabase vector knowledge base.

This n8n workflow runs a Telegram private knowledge-base bot where only admin-approved users can ask questions (text or voice). Answers are generated by an OpenRouter chat model and grounded with retrieval from your Supabase vector store—kept current via a manual Google Docs (Google Drive) to Supabase sync.

What this workflow does

• Telegram trigger: Starts on incoming Telegram messages and callback queries.
• Access control: Admin button callbacks (from the configured control chat) save approved/denied/blocked status into Redis and notify both the admin and the target user.
• User request flow: For user messages, it reads access status from Redis. If access is pending, denied, or blocked, it replies accordingly; otherwise it creates a pending request in Redis and sends an approval request to the admin via inline buttons.
• Approved answering: For approved users, it handles /start and /help, rejects unsupported input, and processes voice messages by downloading the Telegram file and transcribing it with OpenAI.
• Grounded response generation: Sends the user’s question to a LangChain agent using an OpenRouter chat model with Redis chat memory and a Supabase vector-store retrieval tool to produce a grounded Russian answer.
• Telegram-safe delivery: Splits long answers into chunks and sends them back through Telegram.
• Knowledge base updates: Includes a manual trigger to list Google Docs from a specified location and sync them into Supabase.

Use cases

• SaaS operators who need a private Telegram support bot with approval-based access.
• Automation engineers building a knowledge-grounded Q&A system using Supabase vectors.
• Teams who maintain documentation in Google Docs and want periodic updates to an AI retrieval index.

Technical details

• n8n nodes/logic: if, set, code, redis, switch, telegram
• Integrations: Telegram (private bot), Redis (access + chat memory), OpenRouter (chat model), OpenAI (voice transcription), Supabase (vector store retrieval), Google Docs/Google Drive (manual sync)