Storacha Telegram BOT- April 4,2025 - PLDG2

[adielliot37]

Storacha RAG Telegram BOT -April 4 2025 - PLDG 2

Contributors

@adielliot37

Accomplished Tasks

• Integrated the Storacha RAG pipeline into a Telegram bot interface.
• Enabled users to upload knowledge via text, URLs, and PDFs directly from Telegram.
• Implemented a faster, memory-based PDF upload mechanism that streams files directly from memory to IPFS.
• Achieved end-to-end flow: Upload (via Telegram) → Chunk → Embed → Store on Storacha/IPFS → Query using Mistral.

Update

• Built a decentralized RAG Telegram BOT that leverages Storacha on IPFS to store knowledge fragments.
• Improved upload speed using in-memory file handling and a streamlined processing pipeline.
• Text, URL, and PDF inputs are efficiently processed, chunked, embedded, and indexed for semantic search.
• What went well:
  • Smooth integration of Telegram bot functionality with the RAG pipeline.
  • Significant speed improvements with the new memory-based PDF handling.
• What was challenging:
  • Optimizing IPFS retrieval speed across multiple gateways.
  • Managing occasional network timeouts and errors during file uploads.

Demo Videos

Questions

• How can I further optimize the upload and fetching speed from Storacha/IPFS?
• Is there a way to integrate caching or prioritize gateway availability for faster retrieval?
• How can we integrate UCAN directly in the Telegram bot for delegated and shared access between bots?

Next Tasks

• Add UCAN support directly in the Telegram bot for multi-agent delegation and shared knowledge access.
• Explore further speed optimizations in file processing and IPFS uploading.
• Implement advanced caching and pre-fetch strategies for IPFS retrieval.
• Integrate basic authentication and namespacing to organize uploaded knowledge per user/project.
• Develop an automated sync/backfill agent that continuously listens for new knowledge fragments.

[adielliot37]

@jamiechicago312

[jamiechicago312]

What an awesome demo 🤩

Pulling in @alanshaw and @fforbeck to follow up with your questions

• How can I further optimize the upload and fetching speed from Storacha/IPFS?
• Is there a way to integrate caching or prioritize gateway availability for faster retrieval?
• How can we integrate UCAN directly in the Telegram bot for delegated and shared access between bots?