← All projects
n8ngroqopenaimongodb

AI lecture summarizer & Q&A bot

Send a lecture recording to Telegram — get an instant summary, and keep asking questions about it afterward.

Full workflow walkthrough available on request.Get in touch

Problem

Replaying a full lecture recording just to find one specific point takes almost as long as watching it live. There's no quick way to get the key takeaways, let alone go back and ask about something covered halfway through.

Solution

A Telegram bot that takes a voice recording, transcribes it instantly using Groq's Whisper API, and runs it through an AI agent that produces a structured summary. Because the agent keeps persistent chat memory, the conversation doesn't end at the summary — you can keep asking follow-up questions about the lecture, and it remembers the full context.

Architecture

Actual production workflow, built in n8n

Telegram voice message
   → Switch (routes the request)
   → Fetch audio file
   → Compress audio (self-hosted service)
   → Groq Whisper API (transcription)
   → AI Agent (OpenAI model + MongoDB chat memory + structured output parser)
   → Switch (output routing)
   → Document or text summary → delivered via Telegram

+ dedicated error-handling path: merge → format → clear error message back to the user

Results

Turns a full lecture recording into a summary in the time it takes to send a Telegram message — with the ability to keep asking questions afterward.

Tech stack

n8nTelegram Bot APIGroq (Whisper)OpenAIMongoDBGoogle Sheets

Got a process eating hours every week?

Let's talk