Retrieval assistantBuilt

Telegram AI Assistant with Google Drive and Supabase (RAG)

A Telegram assistant that answers from your own documents. Drop a file in Google Drive and it becomes knowledge the assistant can use in chat.

The Telegram knowledge assistant workflow in the n8n editor
The workflow in the n8n editor.

The problem

Sales and support teams lose time hunting for answers in shared folders. A generic chatbot does not know your pricing, policies or product details, and makes things up when asked.

How it works

  1. Ingest documents

    A new TXT or CSV file in Google Drive triggers a flow that reads it, splits it into chunks and embeds each chunk with Gemini.

  2. Store as vectors

    The embeddings are stored in a Supabase vector store (pgvector), so the assistant can search by meaning rather than keywords.

  3. Answer in Telegram

    When someone asks a question in Telegram, the agent retrieves the most relevant passages and answers from them.

Where it fits

  • Sales enablement assistant
  • Support knowledge base
  • Internal onboarding bot

What we would add next

  • PDF and DOCX ingestion
  • Source citations in every answer

Have a sales process that still runs on copy and paste?

Tell us how leads reach your team today. We will map it, show you what to automate first and what it would take to build.