We've been working on a way to turn Telegram channels into an auto-updating knowledge base for AI research tools like NotebookLM. The idea is to get specific, deep answers by tapping into expert discussions without the constant manual effort of exporting and updating data.
This builds on the concept of using AI to structure and analyze information from Telegram channels, effectively "learning" from experts you follow. The goal is an AI that operates on specific, current, and deep knowledge rather than just generalized data. While one approach involves manual periodical exports for the expert’s Telegram channels, we aimed to automate this.
How it Works (MVP with Junction Bot):
Our tool, Junction Bot, has been around since 2017 for Telegram automation (content aggregation, filtering, etc.) and is used by over 17,000 people monthly. We've extended it for this use case:
- Collect from Telegram: Point Junction Bot at expert Telegram channels (public, or private if you connect your account). It can copy existing message history and then automatically fetch new messages. For this AI expert scenario, you'd typically focus on text content.
- Auto-Sync to Google Docs: The bot automatically transfers the collected content into Google Docs. These documents act as a continuously updated knowledge base. If a document hits Google's size limit, a new one is created. - Feed to NotebookLM: Use these Google Docs as sources in NotebookLM. You're providing the LLM with a rich, specific, and current context. You can also add your own internal business documents as sources.
What You Get: An AI assistant that provides qualitatively different, deeper, and more current answers based on the curated expert content. This transforms the AI into a partner that helps extract insights and make decisions based on professionals you trust, automatically staying updated with their latest articles. We use it to do business analysis, feeding it also with internal company documentation.
Current Status & Call for Feedback: This is an MVP for this specific scenario, built within the broader Junction Bot product. It's experimental, and we're keen to hear if this approach is useful.
Disclaimer: Junction Bot is a paid tool (basic plan is €3/mo, full GDocs automation from €12/mo), with a free trial available.
We're exploring this application actively. Please share your thoughts and experiences if you try it!
astuteficus•7mo ago
This builds on the concept of using AI to structure and analyze information from Telegram channels, effectively "learning" from experts you follow. The goal is an AI that operates on specific, current, and deep knowledge rather than just generalized data. While one approach involves manual periodical exports for the expert’s Telegram channels, we aimed to automate this.
How it Works (MVP with Junction Bot): Our tool, Junction Bot, has been around since 2017 for Telegram automation (content aggregation, filtering, etc.) and is used by over 17,000 people monthly. We've extended it for this use case: - Collect from Telegram: Point Junction Bot at expert Telegram channels (public, or private if you connect your account). It can copy existing message history and then automatically fetch new messages. For this AI expert scenario, you'd typically focus on text content. - Auto-Sync to Google Docs: The bot automatically transfers the collected content into Google Docs. These documents act as a continuously updated knowledge base. If a document hits Google's size limit, a new one is created. - Feed to NotebookLM: Use these Google Docs as sources in NotebookLM. You're providing the LLM with a rich, specific, and current context. You can also add your own internal business documents as sources.
What You Get: An AI assistant that provides qualitatively different, deeper, and more current answers based on the curated expert content. This transforms the AI into a partner that helps extract insights and make decisions based on professionals you trust, automatically staying updated with their latest articles. We use it to do business analysis, feeding it also with internal company documentation.
Current Status & Call for Feedback: This is an MVP for this specific scenario, built within the broader Junction Bot product. It's experimental, and we're keen to hear if this approach is useful.
Disclaimer: Junction Bot is a paid tool (basic plan is €3/mo, full GDocs automation from €12/mo), with a free trial available.
We're exploring this application actively. Please share your thoughts and experiences if you try it!