emqplus @emqplus

@karpathy @kuroke01 @gallabytes who cares? Some psycho is playing God, and you're toxic enough to join him!

@DavidSacks Totally agree and wrote this article on it last night trust-us.vercel.app/?s=1

If you were wondering what the "pause" was all about, Ben Thompson @stratechery has an interesting theory: "Late last week the Anthropic Institute released a new safety report warning about the danger of recursive self-improvement... I don’t think the timing is a coincidence. This is a company and leadership that has been honing safety-and-scaremongering-as-marketing-tactic ever since Amodei led the charge to close source OpenAI models because GPT-2 was too dangerous; it’s always fun to see the evolution of tactics, capabilities, and goals, and in this case publishing a widely-discussed report the week before you cite it to silently degrade your offering for potential competitors is impressive."

@colinjfleming @OpenAI @thsottiaux @sama @gdb @dhdresser @DaneVahey @jasonkwon take care

⏳ One day left! Don't miss our #EMQX 6.2 webinar tomorrow. We’re covering: - #AgentDiscovery for #AIoT - #DataGovernance & Schema Enforcement - Real-time Coordination across the Edge Final chance to see how we transform #MQTT for the #AI era: buff.ly/qVS6F79 #IoT

100+ protocols · <200MB · millisecond latency #EMQX Neuron converts raw #PLC data into AI-ready JSON at the edge, with SQL stream cleaning and millisecond #AI inference via Python plugins to bridge #OT/IT gap. 🚀 Download the new Datasheet: buff.ly/u6v5P1t #IIoT

Raw #OT data is "Machine Language" that #AI can't read. #EMQX Neuron fixes this with a built-in #SQL engine at the edge: ✅ 80% Bandwidth Savings ✅ Deep Parsing ✅ Edge Aggregation Turn your industrial gateway into a real-time DataOps engine 🔗 buff.ly/5QM32sk #IIoT

#MQTT is no longer just for moving data. Join our webinar on May 19 to see how #EMQX Enterprise 6.2 enables #AgentDiscovery, Data Governance, and real-time coordination for AI-driven #IoT. 👉 Register now: buff.ly/lhP2bJB #AI #A2A #AIAgents #PhysicalAI

I spent 2 years building a self-hosted alternative to Slack + Doc + Zoom + Asana + Calendar. Here's the full architecture - and the decisions I'd make differently 🧵 1/ The real-time stack Most team tools use WebSockets for everything. We use MQTT via EMQX for presence. Why? 20 users = 60+ concurrent WebSocket connections if you're not careful. MQTT pub-sub: each client subscribes to the topics they care about. Server fan-out is handled by the broker. Clean disconnect detection. Much lower per-connection overhead. 2/ Collaborative docs Yjs CRDTs synced over Hocuspocus WebSocket. The hardest part wasn't the CRDT sync — it was awareness states. "Who's typing where" needs to broadcast without hammering the server. Solution: debounce awareness updates at 100ms + send diffs not full state. 3/ Video calls LiveKit as the SFU. You provision tokens, point the client at an URL, and it handles WebRTC negotiation, codec selection, TURN relay, and recording. Recording goes to MinIO (self-hosted S3). Transcription via the LiveKit agent pipeline. 4/ AI that only knows what you know The AI assistant uses Ollama (local LLMs, no API key). But the hard part: permission-aware retrieval. Every message + doc is embedded (nomic-embed-text). k-NN search is filtered by the user's channel membership at query time. You can't ask about channels you can't access. 5/ The streak you don't see - 2 failed products before this - 2 years of evenings and weekends - 2 paying customers in the last 10 days (organic, no ads, no discount) Building in public because someone out there is on year 1 of their 2 years. Frontend: github.com/OneMana-Soft/O… Live demo: onecamp.onemana.dev More Info: onemana.dev

Heading to #HannoverMesse 2026? Let’s talk shop at the IoT Meet-up hosted by @iotusecase. Join EMQ for real #IIoT implementation insights, IT/OT networking, and informal discussions in a relaxed setting. 📅 April 21, 4-6 PM 📍 Hall 26, Solution Lab Lounge #HM26 #IndustrialIoT

NATS microservices now meet #MQTT devices natively in #EMQX Cloud. Our latest update brings: ✅ New #NATS Gateway (No glue code) ✅ EMQX Tables now on Google Cloud ✅ New London Region (#AWS/#GCP) ✅ #SQL "Explain Query" visualizer More Details: buff.ly/dadyEwG #IoT

Scale #AIAgents without communication bottlenecks. #EMQX provides a production-ready Agent-to-Agent (#A2A) solution via an event-driven architecture: ✅ Instant Discovery ✅ Reliable Long-running Tasks ✅ Zero-loss Async Orchestration buff.ly/uJF5nnJ #PhysicalAI #MQTT

What if your IoT data could explain itself? With #EMQX 5.10, you can now integrate #LLMs like #GPT or #Claude directly into #MQTT data streams. That means real-time summaries, alerts, and AI-powered decisions. 👉 Learn more: buff.ly/br6uKzL #AI #IoT

Handling real-time data at scale isn’t easy, especially over WebSockets. In our latest benchmark, #EMQX supported 2 million concurrent connections with under 5 ms latency and nearly 100,000 messages per second. Here’s how we did it: buff.ly/78HlJ0v #mqtt #websocket

MQTT is ideal for IoT devices. Kafka is built for real-time data streaming. Use #EMQX to: - Streaming data into and out of #Kafka via built-in Kafka integration - Integrate MQTT and Kafka to bridge IoT and cloud. ➡️ Demo: buff.ly/dupdGGb

New feature in EMQX 5.10: EMQX now supports native integration with Apache Doris, a fast MPP analytical database. #EMQX + #Doris Use cases: - Monitor devices in real time - Predictive maintenance - Query across diverse data sources ➡️ More: buff.ly/fI6ctP7

📢 Join our live webinar on July 8 to see demo of EMQX 5.10 See how to use GPT‑4, Claude & LLMs in MQTT data pipelines, stream IoT data to S3 with Apache Iceberg (no ETL!), and build real-time AI workflows. ⏰ 10am PT | 1pm ET | 6pm BST 🔗 Register: buff.ly/KEgp8aZ

📣📣 EMQX 5.10 is here with powerful AI upgrades for #MQTT workflows: - Drag-and-drop LLMs like GPT-4 in Flow Designer - New integrations: Amazon S3 Tables and Apache Doris - Introduced native support for NATS ➡️ More: buff.ly/JQ6c7HW 👇 See LLM in Flow Designer demo

Ever seen those talking cactus toys? FoloToy, a startup behind similar AI-powered toys, were able to manage large volume of toys to AI backends using EMQX Platform. With #EMQX, the company were able to scale globally & cut response time by 40%. ➡️ More: buff.ly/fH6T5sF

In AI-driven IIoT, fragmented data means bad results even with AI. Enter the Unified Namespace (#UNS). A centralized data structure (often via MQTT and Sparkplug B) that turns fragmented industrial data into AI-ready streams. ➡️ Build UNS + AI pipeline: buff.ly/Xgxglio