Brandon Hoefer

@ese @NUSingapore @khayhoe @DrShepherd2013 @ed_hawkins @Prof_Kashwan GIEHP is exactly the vision I’ve had for Mars the past two years — a living nervous system for an entire planet. I just filed the rad-hard datacenter GPU (63/923,108) that makes the Mars version run in seconds instead of hours, all on tiny smallsats. Would love to compare notes — DMs open. 🚀🔴 #sentinelnetwork

@ManOknowledge @RileyRalmuto I haven’t been able to get a true UI demo to show up in a shareable file like you can with Claude using other models. I’m building an app for my local stack. Code and overall project has been designed with Grok/Perplexity/GPT, but the User experience is being done by Claude.

@RileyRalmuto Kinda feels like an advertisement for Claude...anyone know if Grok is capable of doing this too?

slowly but surely, this is going to be the experience and expressed sentiment of virtually every human being with a shred of ambition. and they’ll all start waking up and we’re all going to be watching with popcorn. we’re still so early, chat. somehow.

@FutureJurvetson #SentinelNetwork

@FutureJurvetson Excited by the ai sats. I designed a system for mars! Edge compute in space will open a lot of doors.

T͢E͢R͢A͢F͢A͢B͢ Launch ⌨ Goal: adding a trillion watts of compute, 2x the entire U.S. electricity market, every year. A new 2nm fab to build ~1 billion AI chips/year. Most will be in space. "We have an announcement to make: the most epic chip-building effort by far." — @ElonMusk tonight Live now ► x.com/i/broadcasts/1…

@JesseRank @openhome @menemazarakis Mini PC is on the way — bringing my local router + agent stack fully online next, then scaling nodes through Tailscale. Appreciate the pointer — I’d been overthinking mesh design when the solution was already there.

Hadn't looked at Tailscale before your comment, but after reading up on it the architecture is very close to what I was picturing for the mesh layer. Thank you for the pointer. The persistent memory is still early, but the prototype supports cross-session saves, conversation summarization, and structured context injection at session start and per prompt.

If you want $100k to work on your project, comment your idea below and I’ll help you land this. I’ll be extra energized to help you land it if you’re building an AI agent that can run on @openhome Shoutout @menemazarakis 👌

@SeeThruHead fair point—dynamic per-request routing shines for cloud/multi-model sessions where you want the best/cheapest tool per turn. My router is static at startup: detects hardware once (MPS, RAM, CPU) and locks the single strongest model that fits reliably. Zero overhead per prompt, ideal for offline edge nodes (Pi5, phone mesh) where latency & battery matter more than switching mid-convo. Later I’ll layer tooling on top so even the ‘best’ model gets access to external capabilities without bogging down the device. Different use case, same goal: efficient local inference.

@BrandonHoefer17 @ihtesham2005 doesn't that entirely defeat the purpose of a router which is to use multiple models in the same session for the things they are good/cheap at?

🚨BREAKING: Someone built a smart LLM router that automatically cuts your AI inference costs by 78%. It's called ClawRouter and the numbers are genuinely insane. Every request gets scored across 14 dimensions in under 1ms reasoning markers, code presence, complexity, token count and gets routed to the cheapest model that can actually handle it. Here's what that looks like in practice: "What is 2+2?" → DeepSeek $0.27/M (saved 99%) "Summarize this article" → GPT-4o-mini $0.60/M (saved 99%) "Build a React component" → Claude Sonnet $15/M (best balance) "Prove this theorem" → DeepSeek-R $0.42/M (reasoning) Blended average across a typical workload comes out to $3.17/M. Compare that to $75/M if you're just defaulting everything to Claude Opus. And the payment model is different from anything else out there. No accounts. No API keys. No shared secrets. You generate a wallet, fund it with $5 USDC on Base, and pay per request. That's it. $5 gets you hundreds of requests. 30+ models across OpenAI, Anthropic, Google, DeepSeek, xAI, and Moonshot. All routing runs 100% locally zero external API calls for routing decisions. 100% Opensource. MIT License. Link in comments.

This computer works when the entire internet goes down 🤯 It's called N.O.M.A.D. It gives you Wikipedia, local AI, maps, medical references, and full education courses even when the grid goes down. - Local Ollama AI assistant works fully offline. - Downloadable Wikipedia, maps, and medical guides. - Built-in data analysis tools and local document search. - Runs on a mini PC with a solar panel and battery 100% Open Source.

Just the start! And if you’re a developer that is building agents or OpenClaw, let’s collab. Fill out your address on dev.OpenHome.com Next batch ships in ~2 weeks and we’re selecting for it NOW

We’re just getting started 🔊 The speaker pitches itself: - Local AI - Run your own agents - Amazing sound Easy pitch. Devs, the very long waitlist always has slots for winners: dev.openhome.com (submit a good idea and we’ll prioritize you!)

@JesseRank Filled out my app today! 🤞

@Starcloud_ Was wondering when a constellation would be announced! Congratulations @Starcloud_

🚀

Offline voice agent running locally on my MacBook. Stack:
Mic → Whisper (speech-to-text)
→ Ollama LLM brain
→ Piper voice synthesis Persistent memory reloads on every boot.
No cloud. Fully local. Next:
• S26+ mobile node
• Pi5 edge node
• Hardware-aware model router #LocalAI #OfflineAI #EdgeAI #MeshAI

@co11ateral @three_cube @_aircorridor @DI0256 Yesss! I’m working on something similar!

We have an article on building your own low-cost LTE network. By the way, it’s completely legal in the US hackers-arise.com/sdr-signals-in… @three_cube @_aircorridor @DI0256

🚨 FINLAND HAS SUCCESSFULLY TESTED A SYSTEM THAT SENDS ELECTRICITY THROUGH THE AIR Imagine walking into a room where your phone, laptop, and smartwatch begin charging automatically—no cables, no plugs, no charging pads to align perfectly. This vision of wireless electricity, long confined to science fiction, has taken a major leap toward reality as Finnish researchers successfully test systems that transmit electrical power through the air. Research teams from the University of Helsinki, University of Oulu, and Aalto University, working alongside Finnish tech companies, have demonstrated practical wireless power transmission using multiple innovative approaches including ultrasonic sound waves, electromagnetic fields, and advanced radio frequency systems. While viral social media posts claiming “electricity moves freely through air everywhere in Finland” exaggerate current capabilities, the actual Finnish breakthroughs represent genuine technological milestones with applications ranging from eliminating disposable batteries in sensors to wirelessly charging warehouse robots and potentially powering electric vehicles on the move—marking a significant step toward the clean, cord-free future of electrical distribution. by Miranda Lindale

@seeedstudio @openclaw That’s awesome, I’m working on a similar project, nice work!

We just hooked up the @openclaw agent to a Meshtastic LoRa network! By connecting it directly to a Meshtastic radio mesh network via our new open-source plugin #MeshClaw, your digital twin can now operate 100% off-grid. Build your resilient AI system today! 👇 GitHub Repo: github.com/Seeed-Solution… #OpenClaw #Meshtastic #MeshNetwork #OpenSource #SeeedStudio #theAIHardwarePartner #theIndieCommunicationSystem

@hamza_automates 701

I BUILT A $5,000 AI AUTOMATION FOR PLUMBING COMPANIES And it runs the entire admin side of the business without the owner touching their phone. - It reads new job requests - extracts the details - generates a quote using their pricing rules - proposes available time slots - books the job - sends confirmations automatically After the job is marked complete, it sends the invoice, payment link, and a review request — all without the office staff chasing anything. Same website. Same phone number. Same tools they were already using. Just a system handling the work that usually eats hours every day. If you want to see the full breakdown — how the automation works, the logic behind it, and how to build one yourself: Comment “701” + Like + Repost and I’ll send it over. (must follow for DM)

@JesseRank @openhome @menemazarakis

Developing an AI mesh comms network with customizable local nodes (persistent memory + recall/restore) as POC for my Sentinel Network satellite concept. Filed provisional patent last fall on radiation-tolerant edge GPU for deep-space small sats (US 63/923,108). Mars needs real-time eyes and safety data. Applying to Eigenprize to fund next steps (Pi 5 nodes + radiation testing).