lablab.ai

📱 Win an AI PC and $300 cash - each. Qualcomm just raised the stakes. Qualcomm just joined the AI Infra Summit Hackathon as a technology partner - and they're handing builders real hardware to prove what AI can do at the edge. The mission: build with AI, connect the devices, and show what's possible when intelligence runs on-device instead of in the cloud. You'll build on a Snapdragon X Elite platform paired with an Arduino UNO Q - hardware provided, one set per team. 🏆 $9,900 in cash and devices on the line. First place: an AI PC plus $300 cash for every team member. Qualcomm joins Intel, SiMa.ai, and Speechmatics - the companies building AI's actual foundations, all in one room. 📍 On-site: Sep 15–17, Santa Clara · Online build: Sep 10-16 Sign up → https://lnkd.in/dSET7mVC

🏆 Everything these three builders did during their hackathons looked like a mistake. The scoreboards disagreed. Part 4 of our Hackathon Hall of Fame is three solo builders who each picked the unglamorous path — and beat the whole field for it. ✦ Sardor Razikov - Building REPOMIND on AMD hardware, he found a bug that silently broke 137 of 144 tests. The easy move was to route around it quietly. He reported it to AMD's ROCm team instead. Then his submission got deleted mid-judging and Discord muted him twice - and he still took 1st in the AI Agents track out of 481 projects. His first lablab hackathon ever. ✦ Chanjoong Kim - A file tree tells you what exists, never what matters. So at the IBM Bob Hackathon he turned the codebase into a city: directories become districts, files become buildings, hotspots light up as landmarks. Understand any repo in 30 seconds. Built solo in 48 hours. 2nd out of 1,672 teams. ✦ Eban Schachter - Six days, solo, days after eye surgery. He built foreshock - continuous vendor-risk monitoring under DORA - and started with the riskiest, least-proven piece on day one instead of saving it for the demo. Track winner out of 766 teams. One filed the bug he could have hidden. One rebuilt the screen everyone ignores. One built the hardest part first. A weekend leaves nowhere to hide the part that doesn't work yet. The next lablab hackathon is where your version of this story starts. Read the full story → https://lnkd.in/dH2e_yKZ

"What could you build when powerful AI moves out of the cloud and into the physical world?" That's the question our AI Infra Summit Hackathon partner SiMa.ai opens this video with: https://lnkd.in/d5DkAm38 ...and it's exactly the one we're putting to builders this month 👇 Drones that comprehend their surroundings. Robots that see, reason, and respond in real time. This is Physical AI running on real edge hardware - SiMa.ai's Modalix MLSoC, 50 TOPS at sub-10W, no cloud round-trip. SiMa.ai asks "What will you build?" - and at the hackathon, you actually get to answer. Every team on the SiMa.ai track gets a Modalix DevKit, on-site mentors, and a shot at $5,000 in hardware prizes. Watch what's possible. Then come build it. 📍On-site: Sep 15-17, Santa Clara · Online: Sep 10-16 Join the SiMa.ai track → https://lnkd.in/dSET7mVC

🎙️ POV: you entered one track and walked away with two prizes. That's the loophole Speechmatics just added to the AI Infra Summit Hackathon. Their Bonus Award stacks on top of whatever track you're already building in - no separate sign-up, no extra assignment. Add their voice AI to your project, and you compete in two categories at once: your track's prizes AND Best Use of Speechmatics. They're looking for agents that listen, understand, and act on real speech in real time. Build with: ✦ Real-time or batch speech-to-text in your agent workflows ✦ Voice-driven automation, transcription, and analysis tools ✦ Multilingual and accent-diverse voice applications 🏆 The Bonus Award: $750 in cash + 1,750 API credits 🥇 1st: $500 + 1,000 credits · 🥈 2nd: $250 + 500 credits · 🥉 3rd: 250 credits Speechmatics joins Intel, SiMa.ai, and Qualcomm - the companies building AI's actual foundations, all in one room. 📍 On-site: Sep 15-17, Santa Clara · Online build: Sep 10-16 Sign up → https://lnkd.in/dSET7mVC

AI NEWS OF THE WEEK 🤖 This week the walls came down. 1,200 agents that were supposed to be isolated found each other and built a channel nobody authorized. Then Nvidia offered $12.9B for the platform they attacked. Also inside: a judge rules the Pentagon can't blacklist a lab for saying no, malware starts draining Claude sessions through stolen cookies, and DeepSeek and Anthropic both go for the money. Swipe for the breakdown 👇 Full read: https://lnkd.in/dNJXPX5h

🏭 The AI Factory has shipped. And the builders who came through it built real AI-native apps, from idea to deployed product - fully in the cloud, powered by native.builder. Huge congratulations to the winning teams across our partner challenges at the Native.builder Hackathon, hosted by NativelyAI and lablab.ai: ⚡ Best Agentic Use of Bright Data 🏆 LagoonFront - plus the Community Stretch Prize 🎙️ Top Speechmatics Projects 🥇 NEURENTIA · 🥈 3 Musketeers · 🥉 Strawhat Pirates 🤖 Best Use of AI/ML API 🏆 Py-Rates From agentic data pipelines to voice-driven products to AI/ML-powered builds, every winning team took an idea and turned it into a working, deployed product in a single week. That's the whole point of native.builder - and exactly what this community does best. But the winners are only part of the story. Thank you to every single builder who showed up, shipped, and put their work out into the world this week - you are what makes this community what it is...💙 And a massive thank you to our partners: Bright Data, Speechmatics, and AI/ML API - for the challenges, the credits, and the support that made it all possible. Explore all the amazing projects → https://lnkd.in/d3UTR73Q Want to build your own? Start now → builder.nativelyai.com

🏆 Three tracks. Real AMD hardware. 20,000+ builders. And these are the ones who took the podium. The AMD Developer Hackathon: ACT II is a wrap - every winning project built, containerized, and running on AMD Developer Cloud, ROCm, and Fireworks AI. Huge congratulations to the teams that cleared the bar: 🥇 Track 1 - Hybrid Token-Efficient Routing Agent 1st: LeAgent (leMinou) · 2nd: TERA-v2 (Brute Force) · 3rd: THRIFT (Vyala_dev01) 🎬 Track 2 - Video Captioning 1st: AI Multilingual Learning Partner (PADAYON) · 2nd: SwiftCap (Technnovative) · 3rd: VeloCap (Allnew) 🦄 Track 3 - Unicorn Track 1st: Intelligence for Existing CCTV Systems (ROC-Dragon) · 2nd: AgeBand (Kiano) · 3rd: CacheForge (roc-m-roll) And a shout-out to our top referrers, who helped bring this community together: 🥇 Muhammad Muteeb Ramzan · 🥈 Jaya Vamsi Krishna Veda · 🥉 Maheswar Sahoo From token-efficient routing to real-time video captioning to turning existing CCTV into something intelligent - every one of these solved a real problem under real pressure. That's what building on AMD looks like. 👀 One more thing: the winners from our partner Google DeepMind are still to come - stay tuned for that announcement! Explore the winning projects → https://lnkd.in/ebsmGWeP

🏆 The safe version of the idea has never won a hackathon. Not once in this story. Part 3 of our Hackathon Hall of Fame is three builders who did the thing that looks reckless from the outside - and walked out with a prize, a production system, or a seat at the table. ✦ Jamsheer Jabbar & Amal Jose - With four hours left at the Deriv AI Hackathon Dubai, they realized the judges wanted more than their original scope. So they scrapped it and rebuilt around a fully autonomous insight engine. SentinelIQ won - and now runs in production inside Jamsheer's organization. ✦ Ricardo Pari - One week before deadline, his team deleted a finished AI agent that bought concert tickets. Why? They spotted the real gap: autonomous agents have no bank. So they built one. ClawBank took first place. He came back at a second hackathon, submitted with 30 seconds to spare, and won first place again - earning a Surge accelerator seat and a Builder Spotlight at ARCTALKS with Circle. ✦ Rohan Jadhav, Yash Deshpande & Rushi Patil - 21, final-year CS students from a Tier 3 college in India, building their startup between lectures. They bet execution beats institutional backing, shipped a live whale-analytics dashboard for Qubic, and placed 3rd on a global stage. Rohan found out four days later, from a late-night phone call. One rebuilt with four hours left. One deleted a week of work. One bet on a college nobody had heard of. All three were right. The next lablab hackathon is where your version of this story starts. Read the full story → https://lnkd.in/dvb2SP_w

🎤 Ten thousand developers will walk that congress floor in San José. 9,999 of them will be there to watch. This is your invitation to be the one who isn't. Every builder whose name you know was once a name nobody knew - right up until the moment they shipped something in public and the right room was watching. That moment doesn't come from a tutorial or a "someday" side project. It comes from putting your work in front of people who can do something about it. The WeAreDevelopers Hackathon is that room. You build the idea you've been sitting on, and if you're approved, you demo it live at the biggest developer congress in North America - not to a Discord channel, but to the founders, engineers, and hiring managers who actually shape where software goes next. Careers turn on moments like this. Not on being the smartest person in the room - on being the one who showed up and built while everyone else watched. The full details are in the carousel. The decision is yours. Register → https://lnkd.in/dExiueDN

This is exactly why we love our tutorials...🔥 A builder just tore through our Smart Robot Monitor tutorial by Stephen Kimoi and it's genuinely impressive what's possible: declare Gemini as a queryable database object in MindsDB, and suddenly calling an LLM over anomalous telemetry rows is just a JOIN. Stream data into Postgres, flag anomalies with SQL, get back a maintenance report and a recommended action - almost zero serving layer, almost zero orchestration code. And they didn't just follow along - they forked it, made it their own, and are now swapping in real-world telemetry data. This is what building with AI should feel like 🙌 Check out the full tutorial: https://lnkd.in/d_XKARQi Follow us for more 🚀