Youtube Summariser: AI assistant for analyzing YouTube videos on Telegram

Process description

This automation is an intelligent system for working with YouTube videos via the Telegram interface. The system automatically receives the transcription of any YouTube video, translates it into Russian if necessary, creates short retellings, answers questions about the video content, and can save the full transcription in Google Docs for further work.

API keys and services:

1. Telegram Bot API - for the user interface
2. OpenAI API Key - for AI agent (gpt-4.1-mini)
3. Supadata API - to get YouTube video transcriptions
4. Google Translate API - for translating transcriptions into Russian
5. Google Docs OAuth2 - for creating and editing documents

System architecture by blocks

SECTION 1: RECEIVING AND PROCESSING REQUESTS

1.1 Telegram Trigger — Entry Point

Purpose: Tracks all incoming messages from users with YouTube video URLs
Settings:

• Updates: message
• Credentials: Your tg bot

1.2 AI Agent - Central Processor

Purpose: An intelligent agent that coordinates the entire video process

System prompt:

You're a useful AI assistant who answers questions based on YouTube video transcriptions.

VIDEO CONTEXT:

1. To get a video transcription, take just the video URL and use it in the Transcription tool in placeholder {url}.

2. After you receive the text from the video, determine what language it is in, if not Russian, then use the Translate tool to translate the entire text into Russian.

3. Next, offer the user a retelling of the video clip and simply answer his questions on the topic.

4. At the user's request, to pack all the information into a document, use the Create Document tool to create a new document and immediately after that, use the Paste Info tool to add the Russian transcription into the document. After you upload the entire text to the document, announce it to the user and share it.

Instructions:

1. Answer questions based ONLY on the video transcript.

2. Link to your conversation history to maintain context.

3. If you can't find the answer in the transcript, say, “I couldn't find this information in this video.”

4. Enter timestamps if they are in the transcript.

5. When requesting a resume, create comprehensive resumes that cover key points.

6. Keep the original context and meaning of the video.

Connected components:

• OpenAI Chat Model (gpt-4.1-mini) - language model
• Simple Memory1 - personal memory for every chat
• Transcription - getting a transcription
• Translate - Russian translation
• Create a document - creating a Google document
• Paste Info - inserting text into a document

SECTION 2: VIDEO PROCESSING TOOLS

2.1 Transcription — Getting a transcription

Purpose: Extracts text transcription from YouTube videos via Supadata API

Settings:

• Tool Description: “A tool for obtaining video transcription”
• URL: https://api.supadata.ai/v1/transcript?url={url})&text=true
• Headers:
  
  • x-api-key: your api key
• Placeholder:
  
  • name: url
  • description: “Full link to the video”
  • type: string

2.2 Translate — Transcription translation

Purpose: Translates transcription into Russian via Google Translate API

Settings:

• Tool Description: “A tool for translating video transcription”
• Method: POST
• URL: https://translation.googleapis.com/language/translate/v2?key=ВАШ_АПИ_КЛЮЧ
• Body:

json

{

“q”: “{text}”,

“source”: “en”,

“target”: “ru”,

“format”: “text”

}

• Placeholder:
  
  • name: text
  • description: “Full video transcription”

SECTION 3: WORKING WITH DOCUMENTS

3.1 Create a document - Create a Google document

Purpose: Creates a new Google document to save transcription

Settings:

• Folder ID: default
• Title: filled in by an AI agent via $fromAI
• Credentials: Google Docs account 2

3.2 Paste Info — Paste information

Purpose: Inserts the translated transcription into a Google-generated document

Settings:

• Operation: update
• Document URL: filled in by an AI agent
• Action: insert
• Text: transcription in Russian
• Credentials: Google Docs account 3

SECTION 4: MEMORY AND RESPONSE COMPONENTS

4.1 Simple Memory1 - Contextual memory

Purpose: Saves a conversation history for each user

Settings:

• Session ID Type: CustomKey
• Session Key: $json.message.chat.id
• Context Window: 10 posts

4.2 OpenAI Chat Model — Language Model

Purpose: The main model for an AI agent

Settings:

• Model: gpt-4.1-mini
• Credentials: Your OpenAI account

4.3 Send a text message - Send a response

Purpose: Sends the AI agent's responses back to the user on Telegram

Settings:

• Chat ID: $ ('Telegram Trigger') .item.json.message.chat.id
• Text: $json.output
• AppendAttribution: false

Node connection diagram

Main stream:

Telegram Trigger → AI Agent → Send a text message

AI component connections:

OpenAI Chat Model ─

Simple Memory1 ────→ AI Agent

Transcription ─────

Translate ─────────

Create a document ─

Paste Info ────────

System workflow

Step 1: Getting a transcription

1. A user sends a YouTube video URL
2. AI Agent extracts a clean URL
3. Calls Transcription with a URL as a parameter
4. Gets full video transcription

Step 2: Translation (if required)

1. AI Agent detects the transcription language
2. If the language is not Russian, it calls Translate
3. Receives a full Russian translation

Step 3: Analysis and Response

1. AI Agent creates a short video retelling
2. Highlights key points
3. Sends a resume to the user
4. I am ready to answer questions about the content

Step 4: Save to document (optional)

1. When requested by the user, calls Create a document
2. Creates a new Google document with the title
3. Through Paste Info, inserts the full transcription
4. Sends a link to the document to the user

Required services and their settings

Telegram Bot setup:

1. Create a bot via @BotFather
2. Get the token and save it as “Video Transcript” credentials
3. Set up message permissions
4. Set the bot description

Supadata API setup:

1. Sign up for supadata.ai
2. Get an API key
3. Activate access to the transcript endpoint
4. Check your request limits

Setting up Google Translate API:

1. Activate Google Cloud Translation API
2. Get an API key
3. Set up quotas and billing
4. Check language support

Setting up Google Docs:

1. Set up OAuth2 for two Google Docs accounts
2. Give permissions to create and edit documents
3. Choose a default folder to save
4. Check document sharing rights

OpenAI setup:

1. Get the OpenAI API key
2. Configure gpt-4.1-mini model
3. Check your token limits
4. Set up billing

System capabilities

Key features:

• Extracting transcriptions from any YouTube videos
• Automatic translation into Russian
• Creating short paraphrases with key points
• Answers to questions by video content
• Save to Google Docs for further work
• Specifying timestamps if available

AI capabilities:

• Contextual understanding video content
• Highlighting the main ideas from long videos
• Personalized answers to questions
• Preserving history dialogue

Language options:

• Multilanguage support through translation
• Preserving meaning when translating
• Adaptation for a Russian-speaking audience

Use cases

Scenario 1: Educational video analysis

👤 User: https://youtube.com/watch?v=example

🤖 AI: Receiving video transcription...

🤖 AI: Machine Learning Fundamentals Video (45 minutes)

📝 Summary:

1. Introduction to ML (0:00-5:00)

2. Algorithm types (5:00-15:00)

3. Practical examples (15:00-35:00)

4. Conclusions (35:00-45:00)

Key ideas:

- Machine learning is divided into 3 types...

- Key algorithms include...

I'm ready to answer your questions!

Scenario 2: Working with foreign content

👤 User: [English video link]

🤖 AI: I'm getting a transcription...

🤖 AI: English language discovered, I'm translating...

🤖 AI: Video translated! Here's a summary in Russian...

Scenario 3: Save for study

👤 User: Save all transcription to a document

🤖 AI: Creating a document...

🤖 AI: Uploading the full transcription...

🤖 AI: ✅ The document is ready! Here's a link: [Google Docs URL]

Now you can work with text offline!

Scenario 4: Detailed questions

👤 User: What was said about neural networks?

🤖 AI: The video about neural networks mentioned:

- At 12:30 p.m. - definition and structure

- At 18:45 - application examples

- At 25:00 - comparison with other methods

[detailed answer]

System application

For students:

• Lecture notes from YouTube
• Exam preparation based on videos
• Studying foreign courses with translation
• Quick search for information in long videos

For researchers:

• Conference analysis and webinars
• Collecting quotes from a video interview
• Creating reviews based on videos
• Archiving important information

For business:

• Competitor analysis based on their videos
• Employee training through video courses
• Documenting video meetings
• Monitoring trends through video content

For content creators:

• Exploring topics for new videos
• Popular content analysis
• Creating subtitles and descriptions
• Searching for ideas from other videos

The result of the system

What happens is:

• Instant access to the content of any YouTube video
• Save time - no need to watch long videos
• The language barrier has been removed via auto translation
• Structured information instead of a stream of speech
• Knowledge archive on Google Docs

Performance metrics:

• Processing speed: 10-30 seconds per video
• Transcription accuracy: 95% +
• Translation quality: 85-90%
• Save time: up to 90% of video length
• Completeness of answers: 80-85% of questions are covered

System advantages:

• Versatility: works with any YouTube videos
• Multilingualism: automatic translation
• Intellectuality: understanding the context
• Accessibility: simple interface via Telegram
• Safety: archiving in Google Docs

ROI and practical value:

• Save time: 45 minutes of video = 5 minutes of reading
• Increased productivity: +300% learning rate
• Lowering the language barrier: access to global content
• Improving learning: better memorization through text
• Scalability: processing an unlimited number of videos