News & takes

dTelecom Raises Pre-Seed for Decentralized Voice/Video/STT Infra Tailored to AI Agents on Solana

agents
Team of colleagues engaged in a productive meeting around a whiteboard with idea diagrams

dTelecom Raises Pre-Seed for Decentralized Voice/Video/STT Infra Tailored to AI Agents on Solana

dTelecom just launched the first decentralized x402-powered speech-to-text service on Solana, promising 2-5x lower costs and 2-4x less latency than AWS or Google equivalents.[4] Their dRTC platform handles voice, video, STT, TTS, and translation—already processing 35M+ minutes for 50K+ users while generating revenue.[5][6]

Backed by a $1.2M pre-seed (total ~$2.7M), it's built for AI agents needing real-time comms in a $150B market.[5] CEO Petr Malyukov's posts highlight its edge for devs; X users praise it as a game-changer for agentic voice apps.

GitNexus: Open-Source Tool Turns GitHub Repos into Interactive Knowledge Graphs with AI Querying

GitNexus, a new zero-server open-source tool, parses entire GitHub repos or ZIPs in your browser to build D3.js knowledge graphs of code entities like functions and classes.[7] Powered by embedded KuzuDB, it supports natural language queries via Cypher and an AI agent—all client-side with TypeScript/JS/Python Web Workers.[8][9]

Released last week by abhigyanpatwari, it enables RAG-style code intelligence without backend hassle.[7] X buzz calls it a breakthrough: "Parses your repo in seconds → interactive graph → query like 'show me auth flows'".[9] Ideal for devs auditing large codebases.

What This Means For Your Meetings

These tools spotlight a shift toward smarter knowledge extraction from unstructured content—videos, voice, code—that mirrors what happens in meetings. zproger's deduping hack could supercharge post-meeting video reviews, filtering out repeated points across calls to surface unique action items fast. Pair it with tools like Proudfrog, and your meeting transcripts become a deduplicated knowledge base, saving hours on review.

dTelecom's decentralized STT pushes low-latency voice AI closer to everyday use, perfect for real-time meeting agents that transcribe, summarize, or translate on-chain without Big Tech dependency. GitNexus shows knowledge graphs aren't just for code; imagine applying similar browser-based graphs to your meeting history—query "What did we decide on Q1 budget?" across all calls with speaker links and relations.

Key takeaway: Build your edge by layering these hacks on meeting intel—dedupe redundancy, decentralize capture, graph relationships—to turn scattered discussions into a 10x faster personal knowledge engine.

Sources

  1. https://www.youtube.com/watch?v=6BW4lo7f71I
  2. https://www.reddit.com/r/ObsidianMD/comments/1psc9zf/best_way_to_give_ai_gemini_notebooklm_access_to
  3. https://medium.com/@markgrabe/upload-obsidian-folders-to-notebooklm-bd9aa0bc8ea7
  4. https://blog.dtelecom.org/first-decentralized-x402-powered-speech-to-text-service-cf05016e9133
  5. https://x.com/victoryug26/status/2025572698332549123
  6. https://www.dtelecom.org/dTelecom_whitepaper.pdf
  7. https://github.com/abhigyanpatwari/GitNexus
  8. https://www.sitepoint.com/client-side-rag-building-knowledge-graphs-in-the-browser-with-gitnexus
  9. https://x.com/ihtesham2005/status/2026264341419528643

Get the daily briefing

AI, knowledge graphs, and the future of work — in your inbox every morning.

No spam. Unsubscribe anytime.