Загрузка видео...

Не удалось загрузить видео

Возникла проблема при загрузке этого видео. Это может быть связано с временными проблемами сети или видео может быть недоступно.

На главную

Holy cow, is this thing exciting. esp32-P4: 👍 Pretty powerful microprocessor. 👍 High speed MIPI 480x800(!!) touchscreen display. 👍 High speed MIPI camera. ☹️ Onboard wifi/BT. 👀 $43!! My new leading candidate for a next-gen MicroPython SeedSigner.

Keith Mukai

10,398 subscribers

24,044 просмотров • 5 месяцев назад •via X (Twitter)

Искусство Образование Наука и технологии

Anya Rossi• Live Now

Private livecam show

Комментарии: 17

Фото профиля Mike Tolkachev

Mike Tolkachev5 месяцев назад

@SeedSigner UPDATE: It's really enough just to short-circuit the C10 capacitor with a wire or a solder blob. It will add current leakage through 10k resistor R11 to overall consumption, but it’s just 0.33 mA.

Фото профиля Keith Mukai

Keith Mukai5 месяцев назад

@SeedSigner 🔥

Фото профиля Keith Mukai

Keith Mukai5 месяцев назад

I have not tried to get my code running at all on this thing yet. Was literally just delivered. But OMG, my hopes are high right now!

Фото профиля Keith Mukai

Keith Mukai5 месяцев назад

I'm very confident that someone can figure out how to knock out this C6 wifi chip. It'll probably be enough to just chisel off all the tiny little components (resistors?) on the board on either side of the chip. @skot9000 Brute force ftw? 🤷‍♂️😂

Фото профиля McIntosh - The AI Strategy Show

McIntosh - The AI Strategy Show5 месяцев назад

@SeedSigner That’s exactly the kind of thing I would love @KeithMukai . With my eyes the bigger the better!!

Фото профиля odudex

odudex5 месяцев назад

@SeedSigner 😍

Фото профиля Jayr

Jayr5 месяцев назад

@SeedSigner @BitCoisas check this out, would be cool to port @selfcustodykrux to this hardware huh? Cheaper than the K210 and seems as capable (except the K210 AI capabilities?) for a hardware wallet.

Фото профиля Easy

Easy5 месяцев назад

@SeedSigner 🚀

Фото профиля Mark R Pommrehn

Mark R Pommrehn5 месяцев назад

@SeedSigner Cool! Seems perfect!

Фото профиля infomatec

infomatec5 месяцев назад

@SeedSigner 👀

Фото профиля RetroBreeze

RetroBreeze5 месяцев назад

@SeedSigner It's very powerful for what it is!

Фото профиля Binh

Binh5 месяцев назад

@SeedSigner M5stack has a p4 board without the c6, might want to look into that if u don’t want wifi

Фото профиля Geronimo

Geronimo5 месяцев назад

@SeedSigner @SpecterDIY @Schnuartz 👀👀👀

Фото профиля Ignacio Cristerna 2

Ignacio Cristerna 25 месяцев назад

@SeedSigner //Does it run OpenClaw or Kali Linux?

Фото профиля cCross

cCross5 месяцев назад

@SeedSigner That touchscreen invites quick SeedSigner UI experiments on the side

Фото профиля Horoshi⚡

Horoshi⚡5 месяцев назад

@SeedSigner +#Bitchat 😎

Фото профиля Hugo Quinteiro

Hugo Quinteiro5 месяцев назад

@SeedSigner @coffnix

Похожие видео

This is BLITZ 🏍️ starring myself. I've tested a workflow that combines two of the best AI image editing models right now, Nano Banana and Seedream 4. Using these together feels like a superpower. Seedream 4 for instance generates much more dynamic camera angles BUT Nano Banana does a better job with overall consistency, so utilizing both is super beneficial. I wanted to create a sequence where I'm riding a Cyberbike with an electric booster that overclocks the motor to deliver short, high-risk bursts of extreme speed. The location is a white sand desert. I decided to also test MiniMax Design (H3) start and end frame feature which in my opinion is one of the best since it understands so much about context and camera movement. Start and ends frames let you have much better control over camera movement but where Hailuo truly excels is that it understands speed and camera shake which are essential for any fast paced action sequences. The music was generated with Suno 4.5+ and some additional sound effects like the bike and wind in the desert were sourced from Artlist. Edited with Capcut. There's some more tips and tricks in the thread 👍👍

This is BLITZ 🏍️ starring myself. I've tested a workflow that combines two of the best AI image editing models right now, Nano Banana and Seedream 4. Using these together feels like a superpower. Seedream 4 for instance generates much more dynamic camera angles BUT Nano Banana does a better job with overall consistency, so utilizing both is super beneficial. I wanted to create a sequence where I'm riding a Cyberbike with an electric booster that overclocks the motor to deliver short, high-risk bursts of extreme speed. The location is a white sand desert. I decided to also test MiniMax Design (H3) start and end frame feature which in my opinion is one of the best since it understands so much about context and camera movement. Start and ends frames let you have much better control over camera movement but where Hailuo truly excels is that it understands speed and camera shake which are essential for any fast paced action sequences. The music was generated with Suno 4.5+ and some additional sound effects like the bike and wind in the desert were sourced from Artlist. Edited with Capcut. There's some more tips and tricks in the thread 👍👍

Travis Davids

60,830 просмотров • 1 год назад

🇨🇳 Many people don't know this but Chinese high speed trains are now the fastest in the word and officially faster than even the 🇯🇵 Japanese Shinkansen Chinese trains go 350-400km/h (217-250mph) depending on the model While the Japanese Shinkansen goes 320km/h (200mph) max So I went to try it from Chengdu to Chongqing (just $80 for biz class) This one is the slightly older 350km/h model The business class cabin has 5 big seats that can lay flat with a stewardess dedicated to the cabin who serves you drinks like coffee, water, orange juice etc. I guess if it's a longer trip you probably get food served too The journey was very smooth and very fast, it maintained close to 350km/h at all times and there wasn't many stops or breaks (good) It didn't have WiFi, which is kind of a miss but it seems Chinese just use 5G for everything which worked fine for me without interrruption too I'd say the interior could be a bit more beautiful and modern. I usually say things in Japan look ancient but I think the Shinkansen interior there looks a bit more modern than here The trip took 1 hour and it passed about 300km in that time, roughly equivalent to the width of my country the Netherlands 😂 Anyway solid experience (and cheap!) 👍

🇨🇳 Many people don't know this but Chinese high speed trains are now the fastest in the word and officially faster than even the 🇯🇵 Japanese Shinkansen Chinese trains go 350-400km/h (217-250mph) depending on the model While the Japanese Shinkansen goes 320km/h (200mph) max So I went to try it from Chengdu to Chongqing (just $80 for biz class) This one is the slightly older 350km/h model The business class cabin has 5 big seats that can lay flat with a stewardess dedicated to the cabin who serves you drinks like coffee, water, orange juice etc. I guess if it's a longer trip you probably get food served too The journey was very smooth and very fast, it maintained close to 350km/h at all times and there wasn't many stops or breaks (good) It didn't have WiFi, which is kind of a miss but it seems Chinese just use 5G for everything which worked fine for me without interrruption too I'd say the interior could be a bit more beautiful and modern. I usually say things in Japan look ancient but I think the Shinkansen interior there looks a bit more modern than here The trip took 1 hour and it passed about 300km in that time, roughly equivalent to the width of my country the Netherlands 😂 Anyway solid experience (and cheap!) 👍

@levelsio

2,013,282 просмотров • 9 месяцев назад

🔥HOLY SMOKES! $TAO holders! 🚀 SUBNET 19 (VISION) ON BITTENSOR IS ABSOLUTELY CRUSHING IT! In my 5+ years covering crypto and AI, this is one of the most impressive implementations I've seen. The combination of scale, performance, and decentralization is absolutely next level! 🚀 @namoray_dev @Corcel_X 💨 INSANE Speed Performance: - Llama 3.1 8B: 196.18 tokens/s with +107.23% advantage - Llama 3.1 70B: 124.96 tokens/s with +154.96% advantage - Llama 3.2 3B: 166.69 tokens/s with +21.66% advantage 🔥 Top Tier Model Integration: - Meta-Llama-3-70B & 8B Instruct - FLUX.1-schnell for Text-to-Image - ProteusV0.4-Lightning (Text & Image) - Multiple model variations for redundancy 🔥 What Makes This INSANE: - Complete decentralization - No single point of failure - Multiple model choices for redundancy - Real-time performance tracking - Transparent incentive structure The incentive distribution curve shows a healthy network with: - Strong rewards for top performers - Fair distribution across all participants - Clear path for growth and improvement - Sustainable economic model What's truly MIND-BLOWING is how they've managed to: 1. Scale to millions of operations 2. Maintain high quality across multiple tasks 3. Create a fair, competitive marketplace 4. Build in redundancy and reliability 5. Achieve true decentralization This isn't just another subnet - this is the future of decentralized AI inference happening RIGHT NOW! 🔥 1. MASSIVE Scale & Adoption: - We're seeing 7M+ tokens being processed - 14K+ processing steps being executed - Multiple AI models running simultaneously - Incredible miner participation across the network 2. Revolutionary Task Distribution: - Llama 3.1 70B leading with 20% weighting - Avatar Generation at 15% - Perfectly balanced task distribution for optimal network performance - Multiple specialized tasks including Text-to-Image and Image-to-Image processing 3. Elite Performance Metrics: - Top miners hitting 0.00775 incentive rates - Consistent performance across the network - Impressive scaling from top to bottom performers - Strong incentive curve maintaining network quality 📈 Network Performance: - Consistent upward trend in tokens/s - Quality scores maintaining high levels (>0.9) - Steady improvement in miner performance - Rock-solid network reliability ⚡ Platform Highlights: - Permissionless, serverless architecture - Global network of Always-On GPUs - Instant API access - Full decentralization - Multi-model support with seamless switching What makes this TRULY SPECIAL is the consistent upward trajectory in both speed and quality, while maintaining a decentralized architecture. The performance advantages over industry standards (+154.96% for 70B!) are absolutely mind-blowing! 🚀 This isn't just another AI subnet - it's a glimpse into the future of decentralized AI inference! The combination of speed, reliability, and model variety makes this one of the most impressive implementations in the space! 🔥 📽 Watch Now on YouTube and TikTok: Source 🔗

Andy ττ

11,616 просмотров • 1 год назад

The Agentic Literature Review is Now a Reality. 🚀 I watched a student from King’s College London dismantle a task that used to take weeks. His mission? Deconstruct 47 complex academic papers for his dissertation. The old way: ❌ Endless skimming & highlighting ❌ Messy, unsearchable notes ❌ Citation chaos across 5 different tools The new way? He used a single platform and finished the core analysis in an afternoon. This isn't just another "AI summarizer." ResearchCollab is an AI co-pilot for research. I tested it against the standard "academic grind." The difference was staggering: 🔴 Traditional Workflow: Scattered PDFs, chaotic notes, mental burnout. 🟢 ResearchCollab Workflow: • AI instantly surfaces key insights from 250M+ papers without manual prompting. • Auto-organizes and relates everything personalized to the user. • Generates perfect citations (APA, MLA) in one click. • Brainstorms new research directions you might have missed. 👇 See how it works (No Credit Card Needed): Start your free trial → Why this is a silent revolution for knowledge workers: 1️⃣ Students are cutting literature review time by up to 80%. 2️⃣ Research teams are collaborating in real-time, killing version control nightmares. 3️⃣ The "blank page syndrome" is solved. AI helps you generate outlines and spark unique ideas instantly. The most compelling part? It’s not just about speed. It’s about clarity. ✔️ Finds connections between papers you'd never see. ✔️ Keeps your entire research universe in one searchable place. ✔️ Works 24/7 for less than the cost of your monthly coffee budget. This is the "Copilot" moment for academia and R&D. The barrier to high-quality, organized research has just collapsed. 👉 Support ResearchCollab Product Hunt launch : PS: I've compiled a short guide on "The 5-Day Research Sprint" methodology that this enables. Like 👍 and Comment "Research" and I'll DM you the link.

The Agentic Literature Review is Now a Reality. 🚀 I watched a student from King’s College London dismantle a task that used to take weeks. His mission? Deconstruct 47 complex academic papers for his dissertation. The old way: ❌ Endless skimming & highlighting ❌ Messy, unsearchable notes ❌ Citation chaos across 5 different tools The new way? He used a single platform and finished the core analysis in an afternoon. This isn't just another "AI summarizer." ResearchCollab is an AI co-pilot for research. I tested it against the standard "academic grind." The difference was staggering: 🔴 Traditional Workflow: Scattered PDFs, chaotic notes, mental burnout. 🟢 ResearchCollab Workflow: • AI instantly surfaces key insights from 250M+ papers without manual prompting. • Auto-organizes and relates everything personalized to the user. • Generates perfect citations (APA, MLA) in one click. • Brainstorms new research directions you might have missed. 👇 See how it works (No Credit Card Needed): Start your free trial → Why this is a silent revolution for knowledge workers: 1️⃣ Students are cutting literature review time by up to 80%. 2️⃣ Research teams are collaborating in real-time, killing version control nightmares. 3️⃣ The "blank page syndrome" is solved. AI helps you generate outlines and spark unique ideas instantly. The most compelling part? It’s not just about speed. It’s about clarity. ✔️ Finds connections between papers you'd never see. ✔️ Keeps your entire research universe in one searchable place. ✔️ Works 24/7 for less than the cost of your monthly coffee budget. This is the "Copilot" moment for academia and R&D. The barrier to high-quality, organized research has just collapsed. 👉 Support ResearchCollab Product Hunt launch : PS: I've compiled a short guide on "The 5-Day Research Sprint" methodology that this enables. Like 👍 and Comment "Research" and I'll DM you the link.

Anuj

10,780 просмотров • 10 месяцев назад

Day 08 of building the space game I saw in a dream. 🌙🚀 IT FLIES! The spacecraft I showcased on Day 05 is airworthy, certified and flies under the new realistic physics system I was working day and night for the past few days. ☀️Thanks to astrophysics Professor Fable 5 orchestrating his 4 best PhD Opus 5 colleagues we got: Under the hood: – full 6-DOF rigid-body flight, zero arcade shortcuts – real delta-wing aero: stalls, post-stall drag, the works – fly-by-wire assist so a mouse flies her easily with assist off for the hardcore, oleo landing gear, wheel brakes, speedbrake — you can actually land, an onboard black box records every flight. When physics don't obey, I got the receipts 📼 "Certified" isn't a vibe: 65 automated flight tests green + scripted QC sorties (takeoff, ridge crossing, high-speed decel) flown by the machine before I ever touch the stick. These were also done and orchestrated by the agents under the new skillset that gives them specific controls and camera angles for QC. The other part of this physics system is real weather, wind, lift/drag, shears, turbulence. Yes, I managed to crash several times by underestimating these forces - if high enough, stall recovery was easy, but under 1km, difficult. Skill issue i guess Each part of the spacecraft is programmatic, so you can see the ailerons moving, landing gear retracting, cockpit sliding, even the engine lattices waving. The particle system hasn't been worked on yet, so ignore the random snowflakes going out the engine hahah Another team of agents led by another Fable is currently working on the world generator, so the unfinished island won't be so unfinished soon. Oh, as a bonus, GPT 5.6 Sol built a whole hangar module where we can choose ships and have the tug car haul them to the runway. I'll capture that next time! Next up: a glass cockpit worth flying blind in. Demo video soon. Full avionics!

Day 08 of building the space game I saw in a dream. 🌙🚀 IT FLIES! The spacecraft I showcased on Day 05 is airworthy, certified and flies under the new realistic physics system I was working day and night for the past few days. ☀️Thanks to astrophysics Professor Fable 5 orchestrating his 4 best PhD Opus 5 colleagues we got: Under the hood: – full 6-DOF rigid-body flight, zero arcade shortcuts – real delta-wing aero: stalls, post-stall drag, the works – fly-by-wire assist so a mouse flies her easily with assist off for the hardcore, oleo landing gear, wheel brakes, speedbrake — you can actually land, an onboard black box records every flight. When physics don't obey, I got the receipts 📼 "Certified" isn't a vibe: 65 automated flight tests green + scripted QC sorties (takeoff, ridge crossing, high-speed decel) flown by the machine before I ever touch the stick. These were also done and orchestrated by the agents under the new skillset that gives them specific controls and camera angles for QC. The other part of this physics system is real weather, wind, lift/drag, shears, turbulence. Yes, I managed to crash several times by underestimating these forces - if high enough, stall recovery was easy, but under 1km, difficult. Skill issue i guess Each part of the spacecraft is programmatic, so you can see the ailerons moving, landing gear retracting, cockpit sliding, even the engine lattices waving. The particle system hasn't been worked on yet, so ignore the random snowflakes going out the engine hahah Another team of agents led by another Fable is currently working on the world generator, so the unfinished island won't be so unfinished soon. Oh, as a bonus, GPT 5.6 Sol built a whole hangar module where we can choose ships and have the tug car haul them to the runway. I'll capture that next time! Next up: a glass cockpit worth flying blind in. Demo video soon. Full avionics!

Startracker 🔺

58,166 просмотров • 2 месяцев назад

$Build uncompromising high-performance, high-fidelity apps with React. What you see is not the Apple Messages app, nor is it meant to be 1:1. Rather it serves as an aspirational and real example of UI/UX that is possible. Every detail from the bubble animations to keyboard gestures to scroll view insets and safe areas to parallax timestamps required care and attention. Quality UIs come from caring hard. The little chat bubble tails that morph away when the next message animates in. Timestamps that scroll at a different rate than the bubbles. Peek gestures powered by UIKit that lift a chat bubble, tail and all, with a gentle haptic pulse. The subtle shadow of the text composer and responsively sliding down the keyboard in pixel-perfect concert. The composer shrinking its width to fit iPhone’s rounded bezel. The beautiful Liquid Glass refractions have the level of glassiness chosen by the user (new in iOS 27). Scroll view insets that adjust to fit the home bar and keyboard dynamically, creating a sense of depth as the chat bubbles are slightly visible as they scroll under the keyboard. Keeping the keyboard on screen if the prior screen already presented it. Sliding the keyboard and composer with the screen when the user drags to navigate back. I am excited to explore using the new iPhone Duo view controllers natively with React when the time comes. There is always room for innovation and improvement. This chat demo runs at 120 FPS on iOS 27 even on my phone that’s four versions old now. There are several approaches to high-performance lists. I explored a new Expo scroll view that uses the native gesture systems along with react-native’s VirtualView (codenamed “Fling” to reflect the speed of flinging a scroll view), which Meta’s React team designed to prioritize scroll momentum with React’s scheduler. There are tradeoffs with other scroll strategies for how the scroll views gracefully degrade, and they can be swapped out based on what’s best for your app’s use case. And, by virtue of using modern React, high-priority events like user gestures can preempt lower-priority ones like a network response. Let us dispel any myth that using React requires compromising on beautiful, smooth, user-friendly UI/UX that Uses the Platform. Let us dispel any myth frame-perfect UI coordination is not achievable. The Hermes thread and the UI thread can coordinate when needed, like for controlled text inputs. Let us dispel any myth that React Native’s performance is impacted by a bridge, let alone has a bridge. Bridgeless JSI and the Fabric renderer are here for apps that use the New Architecture. What about Android? The app runs and I’ll share a screenshot from my cloud agent’s emulator. With some attention to detail dedicated to Android I’m confident crafted UX is attainable. I’ve spent my tokens once and have a structural head start everywhere. One more thing! Most of this chat’s application code is written with , , , and other familiar elements. The styling is mostly CSS, including linear-gradient(), background-attachment: fixed, and semantic platform colors that adapt to dark mode. Not just appearance - CSS Animations and CSS Transitions run entirely natively. There are a handful of custom native components written in Swift and Kotlin, like the chat bubbles and a custom CAShapeLayer and UIBezierPath for the tails. Native coding is a first-class feature. The way to think about building beautiful apps this way is React and Native, not React vs. Native. Aspire high and raise your ambitions for what is possible!$

Build uncompromising high-performance, high-fidelity apps with React. What you see is not the Apple Messages app, nor is it meant to be 1:1. Rather it serves as an aspirational and real example of UI/UX that is possible. Every detail from the bubble animations to keyboard gestures to scroll view insets and safe areas to parallax timestamps required care and attention. Quality UIs come from caring hard. The little chat bubble tails that morph away when the next message animates in. Timestamps that scroll at a different rate than the bubbles. Peek gestures powered by UIKit that lift a chat bubble, tail and all, with a gentle haptic pulse. The subtle shadow of the text composer and responsively sliding down the keyboard in pixel-perfect concert. The composer shrinking its width to fit iPhone’s rounded bezel. The beautiful Liquid Glass refractions have the level of glassiness chosen by the user (new in iOS 27). Scroll view insets that adjust to fit the home bar and keyboard dynamically, creating a sense of depth as the chat bubbles are slightly visible as they scroll under the keyboard. Keeping the keyboard on screen if the prior screen already presented it. Sliding the keyboard and composer with the screen when the user drags to navigate back. I am excited to explore using the new iPhone Duo view controllers natively with React when the time comes. There is always room for innovation and improvement. This chat demo runs at 120 FPS on iOS 27 even on my phone that’s four versions old now. There are several approaches to high-performance lists. I explored a new Expo scroll view that uses the native gesture systems along with react-native’s VirtualView (codenamed “Fling” to reflect the speed of flinging a scroll view), which Meta’s React team designed to prioritize scroll momentum with React’s scheduler. There are tradeoffs with other scroll strategies for how the scroll views gracefully degrade, and they can be swapped out based on what’s best for your app’s use case. And, by virtue of using modern React, high-priority events like user gestures can preempt lower-priority ones like a network response. Let us dispel any myth that using React requires compromising on beautiful, smooth, user-friendly UI/UX that Uses the Platform. Let us dispel any myth frame-perfect UI coordination is not achievable. The Hermes thread and the UI thread can coordinate when needed, like for controlled text inputs. Let us dispel any myth that React Native’s performance is impacted by a bridge, let alone has a bridge. Bridgeless JSI and the Fabric renderer are here for apps that use the New Architecture. What about Android? The app runs and I’ll share a screenshot from my cloud agent’s emulator. With some attention to detail dedicated to Android I’m confident crafted UX is attainable. I’ve spent my tokens once and have a structural head start everywhere. One more thing! Most of this chat’s application code is written with , , , and other familiar elements. The styling is mostly CSS, including linear-gradient(), background-attachment: fixed, and semantic platform colors that adapt to dark mode. Not just appearance - CSS Animations and CSS Transitions run entirely natively. There are a handful of custom native components written in Swift and Kotlin, like the chat bubbles and a custom CAShapeLayer and UIBezierPath for the tails. Native coding is a first-class feature. The way to think about building beautiful apps this way is React and Native, not React vs. Native. Aspire high and raise your ambitions for what is possible!

James Ide

139,599 просмотров • 12 дней назад

Okay! WAN 3.0 let's go! Whenever I test a new video model, the first thing I always do is test how well it handles language nuance. And yup, WAN 3.0 did it pretty well. I tested it via Venice And nope, this is not sponsored content. I use my own money. I’ve always used Venice for some of my more erotic video work because it’s uncensored and awesome. As simple as that. 👍😎 The reason i quote this post because people keep asking like heeey where do you generate uncensored stuff. Only two places right now, either Venice or D-zine. So there you go! bUTTTT!!!! If there's more, let me know 👍 I guess I could say that WAN 3.0 is probably the closest model to Seedance 2.0 in terms of quality right now. For my very specific needs, I love making dialogue-driven scenes, and WAN 3.0 is absolutely fine for that. In fact, I have to say it’s much better than MiniMax H3 in some ways. But of course, keep in mind that every model has its own strengths and flaws. Unfortunately, when it comes to very precise, extremely dynamic scenes, like what I usually do with my Sari & Ploy series, especially fight scenes, it’s not quite there yet. The only other problem is that sometimes, if you’re unlucky, WAN 3.0 can produce this weird audio artifact. I don’t really know how to explain it. It’s kind of like that flanger-like sound you sometimes get when generating with Seedance. Yeah, WAN 3.0 can occasionally have the same problem, LOL 😆 But performance-wise, for my use case, especially conversational scenes, candid moments, and the kind of raw camera movement I’ve always loved, WAN 3.0 really delivers. For me personally, this is where things start to get really exciting, especially when I think about series production. I can now be much more strategic about which model I use for each type of scene. For example, I’ll still use Seedance 2.0 and 2.5 mainly for more extreme, dynamic fight scenes, while conversational scenes are something I can comfortably do with WAN 3.0. MiniMax H3, meanwhile, feels more like the experimental option. I personally think H3 is a very unique model. Maybe it’s not as powerful for conversational scenes, but it definitely has its own flavor and can produce things that feel different from the others. But you know what, I can’t help it. I really need to say this LOL. It’s funny how a model like Grok Imagine Video used to be sooo good, and I praised it a lot, but now it feels like an extreme slot machine. At this point, I sometimes feel like it’s mostly being hyped by people who just want to glaze Elon Musk. Somehow, it actually went backwards. 😅 And it’s interesting how models like Seedance 1.5 and the previous WAN versions have improved so much over time. It feels very ironic because I wasn’t even a fan of Seedance back then, and the same goes for WAN. But yeah, at the end of the day is the output quality that matters for me.

Sensitive content

Okay! WAN 3.0 let's go! Whenever I test a new video model, the first thing I always do is test how well it handles language nuance. And yup, WAN 3.0 did it pretty well. I tested it via Venice And nope, this is not sponsored content. I use my own money. I’ve always used Venice for some of my more erotic video work because it’s uncensored and awesome. As simple as that. 👍😎 The reason i quote this post because people keep asking like heeey where do you generate uncensored stuff. Only two places right now, either Venice or D-zine. So there you go! bUTTTT!!!! If there's more, let me know 👍 I guess I could say that WAN 3.0 is probably the closest model to Seedance 2.0 in terms of quality right now. For my very specific needs, I love making dialogue-driven scenes, and WAN 3.0 is absolutely fine for that. In fact, I have to say it’s much better than MiniMax H3 in some ways. But of course, keep in mind that every model has its own strengths and flaws. Unfortunately, when it comes to very precise, extremely dynamic scenes, like what I usually do with my Sari & Ploy series, especially fight scenes, it’s not quite there yet. The only other problem is that sometimes, if you’re unlucky, WAN 3.0 can produce this weird audio artifact. I don’t really know how to explain it. It’s kind of like that flanger-like sound you sometimes get when generating with Seedance. Yeah, WAN 3.0 can occasionally have the same problem, LOL 😆 But performance-wise, for my use case, especially conversational scenes, candid moments, and the kind of raw camera movement I’ve always loved, WAN 3.0 really delivers. For me personally, this is where things start to get really exciting, especially when I think about series production. I can now be much more strategic about which model I use for each type of scene. For example, I’ll still use Seedance 2.0 and 2.5 mainly for more extreme, dynamic fight scenes, while conversational scenes are something I can comfortably do with WAN 3.0. MiniMax H3, meanwhile, feels more like the experimental option. I personally think H3 is a very unique model. Maybe it’s not as powerful for conversational scenes, but it definitely has its own flavor and can produce things that feel different from the others. But you know what, I can’t help it. I really need to say this LOL. It’s funny how a model like Grok Imagine Video used to be sooo good, and I praised it a lot, but now it feels like an extreme slot machine. At this point, I sometimes feel like it’s mostly being hyped by people who just want to glaze Elon Musk. Somehow, it actually went backwards. 😅 And it’s interesting how models like Seedance 1.5 and the previous WAN versions have improved so much over time. It feels very ironic because I wasn’t even a fan of Seedance back then, and the same goes for WAN. But yeah, at the end of the day is the output quality that matters for me.

MXVDXN // DAN

322,187 просмотров • 1 месяц назад

🌋🌊 Massive 7.4 Quake Triggers Tsunami Alert | Mindanao, Philippines | Oct 10 2025 (UTC) A magnitude 7.4 earthquake struck off Santiago, Davao Oriental, near Davao City, shaking much of Mindanao and prompting a Pacific Tsunami Warning Center alert for areas within 300 km of the epicenter. ⸻ What’s New • 🕑 18:43 PHT (10:43 UTC) — quake depth ≈ 36 mi (57 km). • 📍 Epicenter: 07°15′N 126°45′E, ~100 km east of Davao City. • 🌊 PTWC: “Hazardous tsunami waves possible for coasts within 300 km.” Why It Matters • 📊 Major quake near the Philippine Trench—a volatile subduction zone. • 🏛️ PHIVOLCS and local LGUs initiating evacuations along the eastern Mindanao coast. • 🌎 Regional tsunami bulletins issued across the Philippine Sea basin. Effects (Developing) • 👥 Strong shaking felt from Davao to Cagayan de Oro; aftershocks reported. • 🏗️ Power outages and structural checks under way in Tagum and Mati. • 💰 Ports and fishers ordered to move vessels to deeper waters immediately. Gunny’s Take This is a serious event along one of Asia’s most dangerous tectonic zones. Shallow, ocean-trench quakes of this size can generate localized tsunami surges within minutes. If you’re on the eastern Mindanao coast — get to high ground now. Aftershocks and secondary tsunami waves remain possible. What to Watch • ⏰ First-wave arrival: 15–45 minutes after main shock. • 🚚 Evac updates from Davao, Mati, and Tagum coastal barangays. • 🛰️ PTWC + PHIVOLCS bulletins for wave height and aftershock tracking. ⸻ BOTTOM LINE: A powerful 7.4 quake has hit Mindanao — move to higher ground and stay off the coast until official all-clear. Lives depend on speed. 🌊 👇 Like • 💬 Comment • 🔁 Share • 👀 Follow Sources: PTWC | PHIVOLCS | USGS | OSINT Defender | Local Gov Feeds — Gunny 🫡 “Fix the facts. Keep the mission.” #Philippines #Mindanao #Earthquake #TsunamiAlert #Davao #Mati #Tagum #PTWC #PHIVOLCS #BreakingNews #GunnyOSINT

🌋🌊 Massive 7.4 Quake Triggers Tsunami Alert | Mindanao, Philippines | Oct 10 2025 (UTC) A magnitude 7.4 earthquake struck off Santiago, Davao Oriental, near Davao City, shaking much of Mindanao and prompting a Pacific Tsunami Warning Center alert for areas within 300 km of the epicenter. ⸻ What’s New • 🕑 18:43 PHT (10:43 UTC) — quake depth ≈ 36 mi (57 km). • 📍 Epicenter: 07°15′N 126°45′E, ~100 km east of Davao City. • 🌊 PTWC: “Hazardous tsunami waves possible for coasts within 300 km.” Why It Matters • 📊 Major quake near the Philippine Trench—a volatile subduction zone. • 🏛️ PHIVOLCS and local LGUs initiating evacuations along the eastern Mindanao coast. • 🌎 Regional tsunami bulletins issued across the Philippine Sea basin. Effects (Developing) • 👥 Strong shaking felt from Davao to Cagayan de Oro; aftershocks reported. • 🏗️ Power outages and structural checks under way in Tagum and Mati. • 💰 Ports and fishers ordered to move vessels to deeper waters immediately. Gunny’s Take This is a serious event along one of Asia’s most dangerous tectonic zones. Shallow, ocean-trench quakes of this size can generate localized tsunami surges within minutes. If you’re on the eastern Mindanao coast — get to high ground now. Aftershocks and secondary tsunami waves remain possible. What to Watch • ⏰ First-wave arrival: 15–45 minutes after main shock. • 🚚 Evac updates from Davao, Mati, and Tagum coastal barangays. • 🛰️ PTWC + PHIVOLCS bulletins for wave height and aftershock tracking. ⸻ BOTTOM LINE: A powerful 7.4 quake has hit Mindanao — move to higher ground and stay off the coast until official all-clear. Lives depend on speed. 🌊 👇 Like • 💬 Comment • 🔁 Share • 👀 Follow Sources: PTWC | PHIVOLCS | USGS | OSINT Defender | Local Gov Feeds — Gunny 🫡 “Fix the facts. Keep the mission.” #Philippines #Mindanao #Earthquake #TsunamiAlert #Davao #Mati #Tagum #PTWC #PHIVOLCS #BreakingNews #GunnyOSINT

Gunnys Adventures

132,764 просмотров • 11 месяцев назад

One question that's been on my mind for years now is: could we use regular multimodal LLMs not necessarily trained for robotics to do the high level robotics intelligence part that VLAs and WAMs attempt to do? The latest explosion of powerful opensource multi-modal LLMs has, IMO, begun to make this possible due both to intelligence and speed. This is GLM 5.3 Flash, which has vision understanding, but isn't meant to be a VLA/VLM/WAM/robotics model at all, controlling an XGO mini wheeled robot quadruped with an arm & gripper. GLM 5.3F simply has access to the robot's high level SDK for controlling movement, arm joints, open/close gripper...etc. It analyzes the frames from the camera and makes adjustments all on its own to solve the task. Nothing was trained here, nothing fine-tuned for this task. Z AI did not make this model for robots and tbh I think they're surprised this works when I talk to them about it! This also works quite well with DSV4F + a vision capable model like Qwen 3.8 27B. I havent tried JUST Qwen 3.8 27B, but I'm sure it works too. I like the "logic" to be a model that's as fast as possible (but still intelligent). There's also an experimental vision version of DSV4F, I'm confident that'll work too and might even be better bc the full loop might be the fastest of all with this model. An obvious question you might wonder is: well why not use VLA or VLM? The hard part about robotics isn't object detection, that's long solved. This also isn't a solution for gait/locomotion...yet, but I actually don't think this is far away either and I've done some experimentation with LLMs in this space in the past and it does show promise. It might actually already be here for quadrupeds, since you dont need super fast IMU readings to maintain balance. I've also tried many of the larger, more generalist, VLAs that you should be able to use with popular robots and tbh there are just so many edge cases that make things hard and not work. You gotta get the camera, lighting, task, everything *just right* or the demo fails. This is for the actual hard part in robotics right now: intelligence, logic, and planning for all the ways the real world just simply isn't perfect. I've trained VLAs. They're super finicky and you're always running into sim2real issues, especially around the camera. You also have to build the whole training pipeline in a simulator, and, if everything does work, you still just have a robot that does this 1 single thing after weeks of work. If you use teleop, this overcomes the "2real" problem, but now you need to painstakingly collect teleop data, and it's only good at that specific task and that particular robot. There is a growing set of egocentric training data for "general purpose" VLAs and world action models (for humanoid form factors), but I'm really starting to wonder: Why? I think we might just sidestep this whole area of research entirely. I didn't need any training data or special environment to work with this quadruped and arm to do the task I was after. This particular quadruped and arm doesn't even exist in the wild yet really, it's a demo build from a company launching it on kickstarter, so it's not like this robot's data exists in the LLM to any real extent. I think this is cool as heck that this works and I am interested to see just how far I can push it. Also this marks the first time that I've finally got a generalist solution to a task I've been trying to solve ever since I became a dad of twins: pick up toys off the ground. This is a big day!

One question that's been on my mind for years now is: could we use regular multimodal LLMs not necessarily trained for robotics to do the high level robotics intelligence part that VLAs and WAMs attempt to do? The latest explosion of powerful opensource multi-modal LLMs has, IMO, begun to make this possible due both to intelligence and speed. This is GLM 5.3 Flash, which has vision understanding, but isn't meant to be a VLA/VLM/WAM/robotics model at all, controlling an XGO mini wheeled robot quadruped with an arm & gripper. GLM 5.3F simply has access to the robot's high level SDK for controlling movement, arm joints, open/close gripper...etc. It analyzes the frames from the camera and makes adjustments all on its own to solve the task. Nothing was trained here, nothing fine-tuned for this task. Z AI did not make this model for robots and tbh I think they're surprised this works when I talk to them about it! This also works quite well with DSV4F + a vision capable model like Qwen 3.8 27B. I havent tried JUST Qwen 3.8 27B, but I'm sure it works too. I like the "logic" to be a model that's as fast as possible (but still intelligent). There's also an experimental vision version of DSV4F, I'm confident that'll work too and might even be better bc the full loop might be the fastest of all with this model. An obvious question you might wonder is: well why not use VLA or VLM? The hard part about robotics isn't object detection, that's long solved. This also isn't a solution for gait/locomotion...yet, but I actually don't think this is far away either and I've done some experimentation with LLMs in this space in the past and it does show promise. It might actually already be here for quadrupeds, since you dont need super fast IMU readings to maintain balance. I've also tried many of the larger, more generalist, VLAs that you should be able to use with popular robots and tbh there are just so many edge cases that make things hard and not work. You gotta get the camera, lighting, task, everything just right or the demo fails. This is for the actual hard part in robotics right now: intelligence, logic, and planning for all the ways the real world just simply isn't perfect. I've trained VLAs. They're super finicky and you're always running into sim2real issues, especially around the camera. You also have to build the whole training pipeline in a simulator, and, if everything does work, you still just have a robot that does this 1 single thing after weeks of work. If you use teleop, this overcomes the "2real" problem, but now you need to painstakingly collect teleop data, and it's only good at that specific task and that particular robot. There is a growing set of egocentric training data for "general purpose" VLAs and world action models (for humanoid form factors), but I'm really starting to wonder: Why? I think we might just sidestep this whole area of research entirely. I didn't need any training data or special environment to work with this quadruped and arm to do the task I was after. This particular quadruped and arm doesn't even exist in the wild yet really, it's a demo build from a company launching it on kickstarter, so it's not like this robot's data exists in the LLM to any real extent. I think this is cool as heck that this works and I am interested to see just how far I can push it. Also this marks the first time that I've finally got a generalist solution to a task I've been trying to solve ever since I became a dad of twins: pick up toys off the ground. This is a big day!

Harrison Kinsley

53,755 просмотров • 22 дней назад

DeMeco Ryans built an entirely new culture in Houston with one word: SWARM. But it's not just a word - it's an acronym. A simple, memorable, and a powerful way to emphasize what they're looking for. "Swarm mentality is just an acronym for I want to build my team with guys who have a special work ethic and a relentless mindset. And that's where the Swarm mentality came in." S.W.A.R.M. = Special Work Ethic And a Relentless Mindset "That's what the Texan culture will be about - swarming. It's attacking everything." But SWARM isn't just about defense. It's all-encompassing. "It involves teamwork. It involves togetherness. How you attack issues and problems - it's swarming this thing all together. How are you gonna attack a defense or offense? It's those 11 guys out there swarming together on the field." This is what culture looks like when it's built with intention: 1: Get the Right People - Culture starts with who you let in the building. Talent without character is a liability. Ryans' filter is simple: "If it's a guy and I got mixed reports on him...next." 2: Keep It Simple, Play Fast - Complexity doesn't win games. Mastery of the basics does. Your players can't play fast if they're thinking. Simplicity creates speed. 3: Coach The Details, Not The Outcome - Good players can have bad habits. Great coaches see past the result. "In big games...it's not the wow plays. It's who can be consistent, who can not screw it up the most." 4: Be Demanding Without Disrespecting - High standards don't require low respect. You can push hard without tearing down. "You can be demanding without being disrespectful." 5: Let Leaders Be Themselves - Leadership isn't one style. Quiet leaders lead by example. Loud leaders lead by voice. Both work. Ryans tells every player: "I want you to be yourself." Culture isn't a sign on the wall - it's what you emphasize, tolerate, and allow. Special Work Ethic. Relentless Mindset. That's SWARM and the Texans standard. (🎥Pat McAfee)

DeMeco Ryans built an entirely new culture in Houston with one word: SWARM. But it's not just a word - it's an acronym. A simple, memorable, and a powerful way to emphasize what they're looking for. "Swarm mentality is just an acronym for I want to build my team with guys who have a special work ethic and a relentless mindset. And that's where the Swarm mentality came in." S.W.A.R.M. = Special Work Ethic And a Relentless Mindset "That's what the Texan culture will be about - swarming. It's attacking everything." But SWARM isn't just about defense. It's all-encompassing. "It involves teamwork. It involves togetherness. How you attack issues and problems - it's swarming this thing all together. How are you gonna attack a defense or offense? It's those 11 guys out there swarming together on the field." This is what culture looks like when it's built with intention: 1: Get the Right People - Culture starts with who you let in the building. Talent without character is a liability. Ryans' filter is simple: "If it's a guy and I got mixed reports on him...next." 2: Keep It Simple, Play Fast - Complexity doesn't win games. Mastery of the basics does. Your players can't play fast if they're thinking. Simplicity creates speed. 3: Coach The Details, Not The Outcome - Good players can have bad habits. Great coaches see past the result. "In big games...it's not the wow plays. It's who can be consistent, who can not screw it up the most." 4: Be Demanding Without Disrespecting - High standards don't require low respect. You can push hard without tearing down. "You can be demanding without being disrespectful." 5: Let Leaders Be Themselves - Leadership isn't one style. Quiet leaders lead by example. Loud leaders lead by voice. Both work. Ryans tells every player: "I want you to be yourself." Culture isn't a sign on the wall - it's what you emphasize, tolerate, and allow. Special Work Ethic. Relentless Mindset. That's SWARM and the Texans standard. (🎥Pat McAfee)

Coach AJ 🎯 Mental Fitness

69,617 просмотров • 8 месяцев назад

II First Drive : Kia Syros EV II ✅ Had a chance to spend time with the Kia Syros EV & have covered aspects of the exterior, interior and more in the earlier posts. HTX+ 51 kWh 🔋 🏷️ A short test drive taken — key pointers: 👇 👉 Seating / Visibility / Steering ➡️ You sit quite upright, visibility is commanding and the overall view is quite good. The quarter glass adds further to the visibility. ➡️ Steering is adjustable only rake, not reach. The seats 💺 are electrically adjustable for the forward-back position and get a solid feel. ➡️ Drive selector is on the steering column thus takes time to get used to ➡️ Koreans are making some of the best steering — giving solid connect with car ,gets drive modes, as well as paddle shifters, while the switches get good tactile feedback. ORVMS view is good 👍 👉 Drive / Suspension ➡️ Took it on a bad road patch to check the suspension response — it was fine and quite stable. Didn’t get much suspension thuds inside the cabin & we weren’t essentially at high speeds ➡️ Driving overall the Syros EV is better vs the ICE Syros. The battery pack weight adds to stability at high speeds. ➡️ Of course, there is quite a bit of body roll evident coz of tall boy Design ➡️ NVH is fine up to 80–90 km/h; above that, some air noise is noticeable from pillars 👉 Drive Modes / Brakes ➡️ Syros EV gets 3 drive modes — Eco, Normal & Sport. ➡️ They work as desired. Even in Eco mode, power isn’t deficient car is eager to go,instant torque in evs helps the cause ➡️ Steering weights up in Sport mode and the car gives quite a pull in this mode thrill ride ➡️ Braking is quite adequate — it’s sorted here as well. 👍 gets all disc brakes 👉 Regenerative Braking ➡️ Syros EV gets 5(five) regenerative levels, with the top one being the i-Pedal mode. ➡️ This is where the car comes to a complete halt (I-pedal) ➡️ It is offered from the base variant as well. ➡️ Rest other levels are 1, 2, 3 & 0. 👉 Infotainment / Features ➡️ The 30-inch display setup works fine, with plenty of details on display,settings ,it’s flow is good no lags whatsoever ➡️ Music quality with Harman Kardon is good, if not great. ➡️ Infotainment system is quite sleek and works flawlessly. ➡️ The view and flow of the system are quite good. 👉 Interior / Space :: ➡️ The interior is where the Syros EV shines. ➡️ It gets ample space and the sense of space is even enhanced by the panoramic sunroof. ➡️ The rear space & interiors are best among under-4-metre cars. ➡️ The quality, fit & finish is simply segment-leading. 👌 ➡️ Reverse camera quality is quite impressive as well ➡️ Loved the car key as well 👉 Overall :: ➡️ Kia Syros EV seems to be a compelling package, especially the 42 kWh variant ➡️ It will outsell Ice Syros version for sure ,Kia india needs to be complemented for Syros ev as a whole ➡️ 42 kWh The base-to-base-up variants are our recommendations, but the 51 kWh goes expensive for our liking and doesn’t look as compelling at that price. ➕ +ves ➡️ Rear space ➡️ Fit, finish & quality impressive for the segment ➡️ Single-pedal driving ➡️ Korean software ➡️ LTW ➡️ Features packed ➡️ 300–320 km real-world range (42 kWh) & 370-400 (51 kWh) ➡️ Charging speeds > ➖ -ves ➡️ Polarising design ➡️ 51 kWh variants are expensive ➡️ Poor variant distribution ➡️ Okish NVH at high speeds ➡️Hyundai rival arriving soon ➡️ Head and tail lamps close to edges prone to damage Kia India Kia Worldwide Kia Motors India Suhail Gulati - ElecTree Ferrari Rules Vahan Warta (वाहन वार्ता ) Torque India Asheesh Arora AP Sagar EV Geeks of India Rakesh Kumar Tesla Club India®

II First Drive : Kia Syros EV II ✅ Had a chance to spend time with the Kia Syros EV & have covered aspects of the exterior, interior and more in the earlier posts. HTX+ 51 kWh 🔋 🏷️ A short test drive taken — key pointers: 👇 👉 Seating / Visibility / Steering ➡️ You sit quite upright, visibility is commanding and the overall view is quite good. The quarter glass adds further to the visibility. ➡️ Steering is adjustable only rake, not reach. The seats 💺 are electrically adjustable for the forward-back position and get a solid feel. ➡️ Drive selector is on the steering column thus takes time to get used to ➡️ Koreans are making some of the best steering — giving solid connect with car ,gets drive modes, as well as paddle shifters, while the switches get good tactile feedback. ORVMS view is good 👍 👉 Drive / Suspension ➡️ Took it on a bad road patch to check the suspension response — it was fine and quite stable. Didn’t get much suspension thuds inside the cabin & we weren’t essentially at high speeds ➡️ Driving overall the Syros EV is better vs the ICE Syros. The battery pack weight adds to stability at high speeds. ➡️ Of course, there is quite a bit of body roll evident coz of tall boy Design ➡️ NVH is fine up to 80–90 km/h; above that, some air noise is noticeable from pillars 👉 Drive Modes / Brakes ➡️ Syros EV gets 3 drive modes — Eco, Normal & Sport. ➡️ They work as desired. Even in Eco mode, power isn’t deficient car is eager to go,instant torque in evs helps the cause ➡️ Steering weights up in Sport mode and the car gives quite a pull in this mode thrill ride ➡️ Braking is quite adequate — it’s sorted here as well. 👍 gets all disc brakes 👉 Regenerative Braking ➡️ Syros EV gets 5(five) regenerative levels, with the top one being the i-Pedal mode. ➡️ This is where the car comes to a complete halt (I-pedal) ➡️ It is offered from the base variant as well. ➡️ Rest other levels are 1, 2, 3 & 0. 👉 Infotainment / Features ➡️ The 30-inch display setup works fine, with plenty of details on display,settings ,it’s flow is good no lags whatsoever ➡️ Music quality with Harman Kardon is good, if not great. ➡️ Infotainment system is quite sleek and works flawlessly. ➡️ The view and flow of the system are quite good. 👉 Interior / Space :: ➡️ The interior is where the Syros EV shines. ➡️ It gets ample space and the sense of space is even enhanced by the panoramic sunroof. ➡️ The rear space & interiors are best among under-4-metre cars. ➡️ The quality, fit & finish is simply segment-leading. 👌 ➡️ Reverse camera quality is quite impressive as well ➡️ Loved the car key as well 👉 Overall :: ➡️ Kia Syros EV seems to be a compelling package, especially the 42 kWh variant ➡️ It will outsell Ice Syros version for sure ,Kia india needs to be complemented for Syros ev as a whole ➡️ 42 kWh The base-to-base-up variants are our recommendations, but the 51 kWh goes expensive for our liking and doesn’t look as compelling at that price. ➕ +ves ➡️ Rear space ➡️ Fit, finish & quality impressive for the segment ➡️ Single-pedal driving ➡️ Korean software ➡️ LTW ➡️ Features packed ➡️ 300–320 km real-world range (42 kWh) & 370-400 (51 kWh) ➡️ Charging speeds > ➖ -ves ➡️ Polarising design ➡️ 51 kWh variants are expensive ➡️ Poor variant distribution ➡️ Okish NVH at high speeds ➡️Hyundai rival arriving soon ➡️ Head and tail lamps close to edges prone to damage Kia India Kia Worldwide Kia Motors India Suhail Gulati - ElecTree Ferrari Rules Vahan Warta (वाहन वार्ता ) Torque India Asheesh Arora AP Sagar EV Geeks of India Rakesh Kumar Tesla Club India®

Xroaders

11,496 просмотров • 1 месяц назад

My son and I went to the gym late last night to work out some stuff after 3 HS games worth of not hitting the way that he wanted. The session had both a data component and a feel / cue component. As a 17 year old who now has high 60's / low to mid 70's bat speed he has struggled in the last week - specifically against low velocity pitching - to let the ball travel deep enough to get good flush contact. Point being that we had a specific thing to work on. So on the technology side we're using Blast to bat speed, attack angle and VBA, and we're using HitTrax to monitor contact depth in addition to the outcome stuff - exit velocity, launch angle, distance and direction. We also were using two different pitching machines - one offset to the left side, the other offset to the right side. In terms of cues what we ended up with was: - As deep as possible while still being on time - High pitch posture, flat bat path and adjust off if needed But along the way we tried all sorts of stuff - everything from trying to hit a ball to the hot dog cart in oppo foul territory off the left hand off-set machine to trying to find middle of the field ball flight from the right hand offset machine to just going pure let it eat anywhere ball flight off of either machine to score a runner on 2B. In my mind, the best instruction environment for hitting is just a conversation that revolves around: - What were we trying to do vs what did we actually do - What signal can we get about why we did what we did - What adjustments do we need to make to bring our results closer to the intended outcome on the next swing But if the instructor and the player don't even speak the same language...man that conversation is going to be pretty limited in its effectiveness. Because he's 17 we're now able to have these kinds of conversations - with all of these various layers. But the critically important part of the whole thing is that this is an actual two way conversation. Meaning that he's telling me how he feels, we're looking at the data to correlate the feeling to the outcome, we decide on how or if we need to adjust intention for the next swing and we just wash, rinse and repeat that same cycle over and over and over. I wouldn't - and I didn't - coach him this way when he was 12. We used the same tech stack, but the data we paid attention to was a lot more limited. Because that was the level of depth he was able to have a productive conversation about at that time. I'm also exceptionally fortunate that at 17 years old he's still playing and I still get to have these conversations with him.

My son and I went to the gym late last night to work out some stuff after 3 HS games worth of not hitting the way that he wanted. The session had both a data component and a feel / cue component. As a 17 year old who now has high 60's / low to mid 70's bat speed he has struggled in the last week - specifically against low velocity pitching - to let the ball travel deep enough to get good flush contact. Point being that we had a specific thing to work on. So on the technology side we're using Blast to bat speed, attack angle and VBA, and we're using HitTrax to monitor contact depth in addition to the outcome stuff - exit velocity, launch angle, distance and direction. We also were using two different pitching machines - one offset to the left side, the other offset to the right side. In terms of cues what we ended up with was: - As deep as possible while still being on time - High pitch posture, flat bat path and adjust off if needed But along the way we tried all sorts of stuff - everything from trying to hit a ball to the hot dog cart in oppo foul territory off the left hand off-set machine to trying to find middle of the field ball flight from the right hand offset machine to just going pure let it eat anywhere ball flight off of either machine to score a runner on 2B. In my mind, the best instruction environment for hitting is just a conversation that revolves around: - What were we trying to do vs what did we actually do - What signal can we get about why we did what we did - What adjustments do we need to make to bring our results closer to the intended outcome on the next swing But if the instructor and the player don't even speak the same language...man that conversation is going to be pretty limited in its effectiveness. Because he's 17 we're now able to have these kinds of conversations - with all of these various layers. But the critically important part of the whole thing is that this is an actual two way conversation. Meaning that he's telling me how he feels, we're looking at the data to correlate the feeling to the outcome, we decide on how or if we need to adjust intention for the next swing and we just wash, rinse and repeat that same cycle over and over and over. I wouldn't - and I didn't - coach him this way when he was 12. We used the same tech stack, but the data we paid attention to was a lot more limited. Because that was the level of depth he was able to have a productive conversation about at that time. I'm also exceptionally fortunate that at 17 years old he's still playing and I still get to have these conversations with him.

Deven Morgan

72,387 просмотров • 5 месяцев назад

New episode: "The Stubborn Genius of James Dyson" A few notes from the episode: 1. Copying is for losers. 2. They only come to you because you're eccentric. They can get conformity anywhere. 3. Be different and retain total control. 4. Great work is just the vision of a single person, pursued with dogged determination that is nothing less than obsession. 5. The best founders are obsessive, impractical, product-driven enthusiasts. 6. Hire blank slates: Young people who haven’t been ruined by the bad habits of subpar organizations. 7. Be fired by an inner strength and self-belief that is almost impossible to imagine in this feckless age. 8. Vision is just long-term stubbornness. 9. The root principle is to do things your way. It doesn’t matter how other people do it. 10. As long as it works, and it's exciting, people will follow you. 11. Demand difference from what exists. 12. Extreme success comes from doing something no one else is doing. 13. Learning from history is a form of leverage. Have an interest, verging on obsession, with the past. (Dyson literally wrote an encyclopedia on history's greatest inventions) 14. Do not sell a half-finished product. 15. People do not want all-purpose; they want high-tech specificity. 16. Don’t lose the direct connection to your customer. 17. There is no such thing as a quantum leap. There is only dogged persistence — and in the end you make it look like a quantum leap. 18. You simply cannot mix messages when selling something new. Customers can barely handle one great new idea, let alone two, or even several. 19. You want a single message expressed clearly. 20. Appeal to a specific need. Don’t make your product too wide. Narrow it down. 21. You have to think about the incentives of the people you’re selling to. You can’t sell a bagless vacuum cleaner to people that make $500 million selling vacuum bags. 22. No one is ever eager to fix a cash machine that isn’t broken. 23. The more you’re able to control, the better your product— and then your company— will become. 24. Don’t copy the opposition. Don’t worry about market research. Follow your own star. This is what success entrepreneurs do. 25. The opposite of following your own star is following the herd. That is a path to dull conformity. 26. For innovative, intrinsically excellent products, the markets are often larger than you can predict. 27. The entrenched professional is always going to resist far longer than the private consumer. 28. No one ever got an idea starting at a drawing board. 29. Founder led sales is the most powerful tool. 30. The inventive mind knows that there is always further questions to be asked and new discoveries to be made. 31. A clever person doesn’t spend 14 years building 5,127 prototypes of the world’s first cyclonic vacuum cleaner. A determined person does. 32. If you have an idea for a new product, you engineer, prototype, manufacture, market, and sell it yourself. Retain total control. 33. If it’s not beautiful you’re not done. 34. The founder is personally accountable for every product sold. 35. People buy stories. Invest heavily in storytelling. Why your product exists. How it is made. Who made it. 36. Control is more important than money. 37. When you own the whole company all decisions are your own. 38. How bad do you want it? Dyson made a new prototype, every day, for more than 1,000 days. Alone. 39. Perseverance is not cheap. 40. A project will die if the original mule doesn’t stay on it. The self-belief is not there to press through the hard times. 41. Your company should be the most exciting adventure of your life. 42. If you make it, sell it yourself. 43. Only the person with the closest relationship with the product can make a success of it. 44. There's nothing wrong with being persistently dissatisfied or afraid. 45. You need the confidence — and the stupidity — to do things differently. 46. One decent editorial counts for 1,000 advertisements. 47. Difference for the sake of difference and the continuous improvement of products simply for improvement’s sake. 48. Persistent trial and error allows you to wake up one morning, after many, many mornings, with a world-beating product. 49. Seek out originality for its own sake. 50. Companies are built, not made. 51. You are more likely to solve a problem by being unconventional and determined than by being brilliant. 52. Never sacrifice quality for speed. 53. Magic — the unique way a product does what it does — is never to be underestimated. There are 100 more ideas in the episode. I hope you listen to it. 50 years of Dyson's career + 70 hours of reading (and simplifying) and me just ripping through idea after idea at 2x speed for 90 minutes. It will be hard to find a better use of time.

New episode: "The Stubborn Genius of James Dyson" A few notes from the episode: 1. Copying is for losers. 2. They only come to you because you're eccentric. They can get conformity anywhere. 3. Be different and retain total control. 4. Great work is just the vision of a single person, pursued with dogged determination that is nothing less than obsession. 5. The best founders are obsessive, impractical, product-driven enthusiasts. 6. Hire blank slates: Young people who haven’t been ruined by the bad habits of subpar organizations. 7. Be fired by an inner strength and self-belief that is almost impossible to imagine in this feckless age. 8. Vision is just long-term stubbornness. 9. The root principle is to do things your way. It doesn’t matter how other people do it. 10. As long as it works, and it's exciting, people will follow you. 11. Demand difference from what exists. 12. Extreme success comes from doing something no one else is doing. 13. Learning from history is a form of leverage. Have an interest, verging on obsession, with the past. (Dyson literally wrote an encyclopedia on history's greatest inventions) 14. Do not sell a half-finished product. 15. People do not want all-purpose; they want high-tech specificity. 16. Don’t lose the direct connection to your customer. 17. There is no such thing as a quantum leap. There is only dogged persistence — and in the end you make it look like a quantum leap. 18. You simply cannot mix messages when selling something new. Customers can barely handle one great new idea, let alone two, or even several. 19. You want a single message expressed clearly. 20. Appeal to a specific need. Don’t make your product too wide. Narrow it down. 21. You have to think about the incentives of the people you’re selling to. You can’t sell a bagless vacuum cleaner to people that make $500 million selling vacuum bags. 22. No one is ever eager to fix a cash machine that isn’t broken. 23. The more you’re able to control, the better your product— and then your company— will become. 24. Don’t copy the opposition. Don’t worry about market research. Follow your own star. This is what success entrepreneurs do. 25. The opposite of following your own star is following the herd. That is a path to dull conformity. 26. For innovative, intrinsically excellent products, the markets are often larger than you can predict. 27. The entrenched professional is always going to resist far longer than the private consumer. 28. No one ever got an idea starting at a drawing board. 29. Founder led sales is the most powerful tool. 30. The inventive mind knows that there is always further questions to be asked and new discoveries to be made. 31. A clever person doesn’t spend 14 years building 5,127 prototypes of the world’s first cyclonic vacuum cleaner. A determined person does. 32. If you have an idea for a new product, you engineer, prototype, manufacture, market, and sell it yourself. Retain total control. 33. If it’s not beautiful you’re not done. 34. The founder is personally accountable for every product sold. 35. People buy stories. Invest heavily in storytelling. Why your product exists. How it is made. Who made it. 36. Control is more important than money. 37. When you own the whole company all decisions are your own. 38. How bad do you want it? Dyson made a new prototype, every day, for more than 1,000 days. Alone. 39. Perseverance is not cheap. 40. A project will die if the original mule doesn’t stay on it. The self-belief is not there to press through the hard times. 41. Your company should be the most exciting adventure of your life. 42. If you make it, sell it yourself. 43. Only the person with the closest relationship with the product can make a success of it. 44. There's nothing wrong with being persistently dissatisfied or afraid. 45. You need the confidence — and the stupidity — to do things differently. 46. One decent editorial counts for 1,000 advertisements. 47. Difference for the sake of difference and the continuous improvement of products simply for improvement’s sake. 48. Persistent trial and error allows you to wake up one morning, after many, many mornings, with a world-beating product. 49. Seek out originality for its own sake. 50. Companies are built, not made. 51. You are more likely to solve a problem by being unconventional and determined than by being brilliant. 52. Never sacrifice quality for speed. 53. Magic — the unique way a product does what it does — is never to be underestimated. There are 100 more ideas in the episode. I hope you listen to it. 50 years of Dyson's career + 70 hours of reading (and simplifying) and me just ripping through idea after idea at 2x speed for 90 minutes. It will be hard to find a better use of time.

David Senra

195,128 просмотров • 1 год назад

✨New demo: what if vibe coding felt more visual? Brian Lovin Mary Rose Cook and I did a game jam using Notion as our "IDE": launching Cursor agents from a task board, and making a custom image for each task 😎 The demo shows 3 ideas for the future of agents: 1) Agents should collaborate across apps. Each app has its focus--Notion AI is good at drafting specs and organizing tasks; Cursor is good at coding. So let them specialize! Today we're launching a new integration where Notion AI can kick off Cursor Cloud Agents to do coding tasks. The Cursor API accepts natural language prompts, so I think of this as "cross-app sub-agents" -- it's kinda cute how it resembles humans hiring outside contractors 😊 BTW: the parallelism of cloud agents is incredibly freeing for creativity, but it also creates a new problem: sooo much work to keep track of! Which brings us to the next idea... 2) Agent orchestration is a data visualization problem. A powerful frame for designing agent UIs is to think of the chat transcripts as the "raw data" and ask: what visual projections might help people make sense of this data at scale? We need to engage our human GPUs -- our visual processing -- to understand what the computer GPUs are doing for us! One thing we can do is use AI to populate traditional UIs like progress bars and status updates. But there are also new possibilities now... For example: when you have a lot going on, it can be hard to identify tasks just by text titles. So we tried generating an AI image for each task -- turns out this helps a lot by giving it a unique visual identity! And of course, it also just makes it super fun to build with friends 😃 Speaking of friends... 3) The future of coding is collaborative. Sometimes it feels like IC engineers are being reduced to middle managers: shuffling information between the team's context and the coding agents that they individually manage. The solution: bring all the people and agents into one shared space, with shared context and visibility! In the video you can get a glimpse of how this feels. Mary, Brian and I record ourselves chatting about ideas, and then we use AI to turn that conversation into a list of tasks on a shared board. As the ideas get built in parallel, we can all monitor progress and review the work together, nothing is siloed. My main takeaway from this game jam was: damn, creativity with friends, at the speed of conversation, is incredibly fun. --- Our goal here is to let anyone use Notion as a fun and creative "software factory" to build software together with your team. Give the Cursor integration a shot and let us know what you think! (AI Image gen in Notion isn't GA yet, but coming soon and already out to some users) And let me know if you'd want a template or more detailed instructions on the setup we showed in this demo...

✨New demo: what if vibe coding felt more visual? Brian Lovin Mary Rose Cook and I did a game jam using Notion as our "IDE": launching Cursor agents from a task board, and making a custom image for each task 😎 The demo shows 3 ideas for the future of agents: 1) Agents should collaborate across apps. Each app has its focus--Notion AI is good at drafting specs and organizing tasks; Cursor is good at coding. So let them specialize! Today we're launching a new integration where Notion AI can kick off Cursor Cloud Agents to do coding tasks. The Cursor API accepts natural language prompts, so I think of this as "cross-app sub-agents" -- it's kinda cute how it resembles humans hiring outside contractors 😊 BTW: the parallelism of cloud agents is incredibly freeing for creativity, but it also creates a new problem: sooo much work to keep track of! Which brings us to the next idea... 2) Agent orchestration is a data visualization problem. A powerful frame for designing agent UIs is to think of the chat transcripts as the "raw data" and ask: what visual projections might help people make sense of this data at scale? We need to engage our human GPUs -- our visual processing -- to understand what the computer GPUs are doing for us! One thing we can do is use AI to populate traditional UIs like progress bars and status updates. But there are also new possibilities now... For example: when you have a lot going on, it can be hard to identify tasks just by text titles. So we tried generating an AI image for each task -- turns out this helps a lot by giving it a unique visual identity! And of course, it also just makes it super fun to build with friends 😃 Speaking of friends... 3) The future of coding is collaborative. Sometimes it feels like IC engineers are being reduced to middle managers: shuffling information between the team's context and the coding agents that they individually manage. The solution: bring all the people and agents into one shared space, with shared context and visibility! In the video you can get a glimpse of how this feels. Mary, Brian and I record ourselves chatting about ideas, and then we use AI to turn that conversation into a list of tasks on a shared board. As the ideas get built in parallel, we can all monitor progress and review the work together, nothing is siloed. My main takeaway from this game jam was: damn, creativity with friends, at the speed of conversation, is incredibly fun. --- Our goal here is to let anyone use Notion as a fun and creative "software factory" to build software together with your team. Give the Cursor integration a shot and let us know what you think! (AI Image gen in Notion isn't GA yet, but coming soon and already out to some users) And let me know if you'd want a template or more detailed instructions on the setup we showed in this demo...

Geoffrey Litt

89,581 просмотров • 6 месяцев назад

The state of LOW-FI and platform support (Death of PCVR): When the Kickstarter ran back in , the main focus was "Next-Gen VR", meaning at the time PCVR, with high hopes for PSVR2 (which hadn't been officially announced). During the next year or so, we (myself, Noah Rayburn and Mark Schramm #VR #AR #MR) focused purely on supporting PC VR hardware. Rift/S, Vive, Index (including controllers), all WMR devices, etc... Mark even added HOTAS and Leap Motion support. BUT, as I'm sure you know, PCVR has been in decline since around the launch of Half-Life: Alyx, which Anton Hand so eloquently called the punctuation mark on the gravestone of PCVR. The Rift store is a ghost town.. like officially.. even according to Meta, Viveport (lets not even go there), what else we got? WMR Store? Was that even a thing? Anyway... Lucky for us, PSVR2 was just over the horizon and we had the dev kits! How well is PSV2 doing though? Guess is (based on speaking with other top VR devs), not that great... Somewhere in there, the great and powerful Noah Rayburn was able to implement a pretty slick "non-VR mode" upon my request to be able to test the game without having to keep taking the HMD off and on while balancing controllers on my knees and working with janky Quest link stuff. Good news! Non-VR mode is great! No just for testing, but for playing. We decided to go full in on it's development and offer it to the consumers. Bonus, it also runs on Steam Deck! (See video below). Now as we gear up to finish the game and run out of money (hopefully in that order), I'm thinking a lot about the market for the game and our placement in it. We have a decent amount of wishlists (66k on Steam and 24k on PS5) so far. Hope is that launching as a PSVR2 title will capture a good chunk of those users, pushing it up on the general PSN store ranking (notoriously hard for indies to do otherwise), and we'll make some sales to traditional flat PS5 gamers as well. Snowballing a bit if we're lucky. Steam launch could do decently well if timed properly, with hopes of the same "VR first" snowballing effect. Why am I telling you all of this? I just thought it would be good to get it in writing and have a place to point. We are starting to run low on money (PCVR presales have basically dried up), so it's also to let you know that the fire is lit under my ass to ship this thing. That said, it won't ship until its good and done. A delayed game is eventually good, but a rushed game is forever bad. Oh, and before you say "Why not port it to Quest", put down the pipe. I understand that you're not a developer, but take my word for it, that's not happening. Not with this budget(or lack thereof), this sized team , or this generation of devices. Ok. that's it. My longest twitter post. PC/PCVR Preorder and early access: Steam Wishlist PSN Wishlist ❤️🧡💙🖤

The state of LOW-FI and platform support (Death of PCVR): When the Kickstarter ran back in , the main focus was "Next-Gen VR", meaning at the time PCVR, with high hopes for PSVR2 (which hadn't been officially announced). During the next year or so, we (myself, Noah Rayburn and Mark Schramm #VR #AR #MR) focused purely on supporting PC VR hardware. Rift/S, Vive, Index (including controllers), all WMR devices, etc... Mark even added HOTAS and Leap Motion support. BUT, as I'm sure you know, PCVR has been in decline since around the launch of Half-Life: Alyx, which Anton Hand so eloquently called the punctuation mark on the gravestone of PCVR. The Rift store is a ghost town.. like officially.. even according to Meta, Viveport (lets not even go there), what else we got? WMR Store? Was that even a thing? Anyway... Lucky for us, PSVR2 was just over the horizon and we had the dev kits! How well is PSV2 doing though? Guess is (based on speaking with other top VR devs), not that great... Somewhere in there, the great and powerful Noah Rayburn was able to implement a pretty slick "non-VR mode" upon my request to be able to test the game without having to keep taking the HMD off and on while balancing controllers on my knees and working with janky Quest link stuff. Good news! Non-VR mode is great! No just for testing, but for playing. We decided to go full in on it's development and offer it to the consumers. Bonus, it also runs on Steam Deck! (See video below). Now as we gear up to finish the game and run out of money (hopefully in that order), I'm thinking a lot about the market for the game and our placement in it. We have a decent amount of wishlists (66k on Steam and 24k on PS5) so far. Hope is that launching as a PSVR2 title will capture a good chunk of those users, pushing it up on the general PSN store ranking (notoriously hard for indies to do otherwise), and we'll make some sales to traditional flat PS5 gamers as well. Snowballing a bit if we're lucky. Steam launch could do decently well if timed properly, with hopes of the same "VR first" snowballing effect. Why am I telling you all of this? I just thought it would be good to get it in writing and have a place to point. We are starting to run low on money (PCVR presales have basically dried up), so it's also to let you know that the fire is lit under my ass to ship this thing. That said, it won't ship until its good and done. A delayed game is eventually good, but a rushed game is forever bad. Oh, and before you say "Why not port it to Quest", put down the pipe. I understand that you're not a developer, but take my word for it, that's not happening. Not with this budget(or lack thereof), this sized team , or this generation of devices. Ok. that's it. My longest twitter post. PC/PCVR Preorder and early access: Steam Wishlist PSN Wishlist ❤️🧡💙🖤

Blair Renaud // LOW-FI 🟥🟧⬛

69,789 просмотров • 2 лет назад

YESTERDAY WAS THAT DAY 😊😊😊 A DAY WITH THE Bolajiogunmolatv 🫶🏽 Have your fav decided to fly all the way just to surprise you? NOOO! So shift back 🤣🤭 I was shocked. It’s very hard to surprise me cos I will always pick up a clue or my instinct will kick in but this time I was legit shocked.. And the fact that she didn’t post in real time for 12 hours so I won’t suspect sealed everything. I was shaking for more than 5 mins. Like how? What! My head wan blow 🤯🤯🤯 When she said “I just came to see you. I came to Abuja for you”… I thought she was whining me but this woman came to my place straight from the airport, spent the day with me and my sisters and left first thing this morning. She didn’t even tell her Abuja friends she was coming just so we can enjoy the day together 🥹🥹. Don’t tell me not to feel special, it’s too late. I AM ALREADY FEELING EXTREMELY SPECIAL 🥰 MY HEAD WAN BURST 🥹🤣 -She came with her brother and was intentional about coming to help me Make content for my brand. His work rate >>>>.. In her words “you gave me your time and that is very important to me”… if you know me, you will know the easiest way to finish me is by supporting my brand 😭🙏🏽🫶🏽 -I have been asking her to send in her measurements for months. Well she came for physical measurements. 💅🏽 -She is such a vibe and holds the best conversation. We were gisting for hours like we knew each other for years.😊 -I got sneak peak of our next YouTube movie and y’all better be ready. Another banger dropping soon 🎉🎉 -Oh! And did I tell you how KIND she is. KIND, GRACEFUL,THOUGHTFUL,HER PRESENCE IS SO SOOTHING AND HER MIND IS BRILLIANT.. She came bearing gifts too. I didn’t even open it cos I was too excited to see her. Will go and open it later today… -SHE IS SOOO PRETTY. OMG! Camera is not doing any justice abeg. Too beautiful and portable. TV definitely makes them look bigger. -Gave her a tour of our 80% done fashion space and she wanted to use hype to finish me. Advised, prayed and encouraged me. -Then we went to watch AJOSEPO the gathering for the 5th time with her. She brought my own personal premiere to me 💅🏽 -Funfact; This was something she didn’t have to do but she did just to show me love. She already messaged me to thank me several times and even chatted on video call and it was very okay. I was always telling her the “Thank you” was enough so you see how this was such a big deal. THANK YOU SO MUCH WIFEY. Yesterday was super special because of you 🫶🏽.. One of the edits her amazing brother did on the spot. 👇🏽👇🏽. He is too good 👍

Ayeesha

82,470 просмотров • 3 месяцев назад

A world first: a demonstration of AI running *on* blockchain as a smart contract 🧠⚡️ The Internet Computer is used – the world's first 3rd gen. blockchain #ICP. AI will become the beating heart of our web3, multi-chain world, and this is only my first demo. Code will follow shortly. This is running on DFINITY's Internet Computer testnet, but you'll be able to take the code and run it on the public network as the NNS is expected to up the per-transaction instruction limit in the coming days. Some important notes. The inference engine used has not been optimized, and we will show vastly greater efficiency in subsequent demos, where the AI runs even faster, and consumes less gas/cycles. We will also propose to the NNS that smart contracts have access to SIMD instructions – which we have determined are deterministic – unlocking vastly more speed and efficiency. Lastly, currently the actor smart contracts ICP hosts run inside a 32 bit environment, which limits their main memory to 4GB. Within the next couple of months, we expect the Internet Computer to move to a 64 bit environment, allowing smart contract memory limits to be raised much higher – allowing for models with far more parameters, and thus power, to be run. Those of you who have followed my posts know that hardware optimization support is planned. Firstly this will involve WASM smart contracts shipping out matrices for processing by the CPUs on existing node machines (another advantage of the Internet Computer running on sovereign node hardware). Secondly, we plan to propose a new public node machine specification, which node providers can build to, for use in subnets specialized for hosting AI smart contracts, in which each machine will incorporate several GPUs. Naturally, this will be packaged with other technological developments that ensure smart contract determinism. This is part of our mission to enable powerful LLMs to run as smart contracts on the Internet Computer – in a forms that are tamperproof, unstoppable, and optionally autonomous (including under the exclusive control of DAOs). Our vision: you will be able to have a chat with a smart contract. A smart contract will be able to coordinate your organization (see my earlier tweet about a "Delphi"). ICP smart contracts will be able to audit Ethereum smart contracts, kitemarking those without backdoors and reentrancy bugs. And smart contracts will be able to do KYC autonmously, matching faces to driving licenses, and more. We will deliver new SDK enhancements, and work with partners, to turbocharge crypto AI developers. Secure and unstoppable AI and third generation blockchain will be two sides of the same coin. Security and AI will be indivisible. AI models will be traded as NFTs. Thanks to Internet Computer's chain key (trustless multi-chain) capabilities, all blockchains will be able to leverage AI smart contracts. The future is beyond exciting. Thanks for watching. Can't wait to give you more demos!

A world first: a demonstration of AI running on blockchain as a smart contract 🧠⚡️ The Internet Computer is used – the world's first 3rd gen. blockchain #ICP. AI will become the beating heart of our web3, multi-chain world, and this is only my first demo. Code will follow shortly. This is running on DFINITY's Internet Computer testnet, but you'll be able to take the code and run it on the public network as the NNS is expected to up the per-transaction instruction limit in the coming days. Some important notes. The inference engine used has not been optimized, and we will show vastly greater efficiency in subsequent demos, where the AI runs even faster, and consumes less gas/cycles. We will also propose to the NNS that smart contracts have access to SIMD instructions – which we have determined are deterministic – unlocking vastly more speed and efficiency. Lastly, currently the actor smart contracts ICP hosts run inside a 32 bit environment, which limits their main memory to 4GB. Within the next couple of months, we expect the Internet Computer to move to a 64 bit environment, allowing smart contract memory limits to be raised much higher – allowing for models with far more parameters, and thus power, to be run. Those of you who have followed my posts know that hardware optimization support is planned. Firstly this will involve WASM smart contracts shipping out matrices for processing by the CPUs on existing node machines (another advantage of the Internet Computer running on sovereign node hardware). Secondly, we plan to propose a new public node machine specification, which node providers can build to, for use in subnets specialized for hosting AI smart contracts, in which each machine will incorporate several GPUs. Naturally, this will be packaged with other technological developments that ensure smart contract determinism. This is part of our mission to enable powerful LLMs to run as smart contracts on the Internet Computer – in a forms that are tamperproof, unstoppable, and optionally autonomous (including under the exclusive control of DAOs). Our vision: you will be able to have a chat with a smart contract. A smart contract will be able to coordinate your organization (see my earlier tweet about a "Delphi"). ICP smart contracts will be able to audit Ethereum smart contracts, kitemarking those without backdoors and reentrancy bugs. And smart contracts will be able to do KYC autonmously, matching faces to driving licenses, and more. We will deliver new SDK enhancements, and work with partners, to turbocharge crypto AI developers. Secure and unstoppable AI and third generation blockchain will be two sides of the same coin. Security and AI will be indivisible. AI models will be traded as NFTs. Thanks to Internet Computer's chain key (trustless multi-chain) capabilities, all blockchains will be able to leverage AI smart contracts. The future is beyond exciting. Thanks for watching. Can't wait to give you more demos!

dom | icp

1,363,716 просмотров • 2 лет назад

Dwarkesh is WRONG about the "Output Gap" The narrative around Artificial Intelligence has shifted perceptibly in late 2025. After years of exponential hype, a sense of disillusionment has begun to settle over the industry. Commentators and analysts, most notably podcaster and writer Dwarkesh Patel, have recently highlighted what is being called the “Output Gap.” This is the uncomfortable discrepancy between our models’ skyrocketing performance on benchmarks and the relatively stagnant growth in macroeconomic productivity. We have reached “superhuman” capability on tests, yet global GDP hasn’t skyrocketed, and the promised revolution feels curiously delayed. This frustration stems from a fundamental misunderstanding of where we are in the technology cycle. The industry is currently fixated on “Day 0” capabilities—the raw intelligence of the models, the scaling laws, and the saturation of academic benchmarks. However, the bottleneck has shifted. We are no longer limited by the intelligence of the model, but by the inertia of the enterprise. The “Output Gap” is not a failure of technology; it is a lag in organizational digestion. We have invented a powerful jet engine, but we are essentially frustrated that it hasn’t revolutionized travel before we’ve even built the airframe to mount it on. The primary fallacy driving this disappointment is the expectation of the “drop-in remote worker.” Many observers equate Artificial General Intelligence (AGI) with a digital human that can be onboarded, culturally assimilated, and left to run autonomously with minimal supervision. Because current agents cannot seamlessly replace a human employee in this one-to-one fashion, the conclusion is often that the technology has stalled. This view misses the forest for the trees. Disruptive technologies rarely act as direct replacements; instead, they require a complete restructuring of how work is done. In reality, the barrier to adoption is what IT professionals call “Day 2 Operations.” Day 0 is the exciting launch; Day 2 is the boring, messy reality of governance, security, and maintenance. For an enterprise to deploy an autonomous agent, it isn’t enough for the model to be smart. The organization must solve for Role-Based Access Control (RBAC), SOC 2 compliance, liability frameworks, and data sovereignty. Right now, most organizations lack the infrastructure to handle “non-human” identities that have access to sensitive corporate data. Consider the security implications. A human employee has physical limitations and a single identity. An AI agent is an “always-on” entity that, if given improper permissions, could theoretically read every email in a company server to “optimize workflow.” Security teams (CISOs) are rightly terrified of this prospect. Until we develop granular access controls specifically designed for agents—effectively an “RBAC for AI”—the widespread deployment of autonomous agents will remain blocked by the “Departments of No”: Legal, HR, and Security. This operational friction explains why we are seeing a massive divergence between individual and enterprise adoption. Individually, adoption is rampant; we are in the “Shadow IT” era of AI, where employees secretly use tools like ChatGPT to boost their personal productivity. However, at the organizational level, adoption is glacial because the institution’s primary mandate is risk management, not speed. The C-suite is asking questions about ROI, liability, and data leakage that the current software ecosystem cannot yet answer satisfactorily. History offers a comforting precedent for this timeline. We are currently effectively in the “2002 era” of virtualization. In the early 2000s, virtualization technology (like VMware) was technically viable, but it took nearly a decade to become the default enterprise standard. It required years of maturing management software, security protocols, and cultural shifts before “The Cloud” became a reality. AI is undergoing the same digestion period. The technology is ready, but the enterprise “rails” required to run it are still being laid. This leads us to the “Mechanical Horse” fallacy. When the automobile was invented, people didn’t need a mechanical horse that walked on four legs; they needed a car. But the car required asphalt roads, not dirt paths. Similarly, we are currently trying to force AI into human-shaped workflows (”jobs”) rather than rebuilding the workflows to suit the AI. We are waiting for the mechanical horse, failing to realize that the true disruption comes from unbundling the work entirely. A “job” is essentially a bundle of tasks, context, and responsibilities aggregated for a single human. AI does not replace jobs; it unbundles tasks. The transition we are navigating involves breaking down these bundles and determining which specific value streams can be automated by agents. This requires task decomposition and new management frameworks—a “Council of AIs” approach—rather than a single omnipotent bot. This restructuring is an organizational physics problem, not a computer science problem. The disconnect is further exacerbated by the differing timelines of developers versus executives. Developers and researchers live in a world of root access and rapid iteration, often blinded to the molasses-like speed of corporate change management. A CFO, conversely, thinks in fiscal quarters and compliance audits. They require proven ROI and standardized best practices before signing off on widespread automation. Currently, there are no industry standards for deploying autonomous agents, which halts most conversations at the boardroom door. Therefore, the “stalling” progress is an illusion caused by looking at the wrong metrics. If you look at model cards and benchmarks, progress is linear and fast. If you look at widespread economic integration, the curve is flat. This S-curve of adoption is always significantly behind the S-curve of capability. We are in the flat part of the adoption curve, characterized by high hype, high friction, and frantic infrastructure building behind the scenes. The next three to five years will likely be dominated not by flashy new model capabilities, but by the “boring” work of integration. We will see the rise of startups and consultancies dedicated solely to “AI Governance,” “Agent Identity Management,” and “Context Orchestration.” These are the boring rails that will eventually allow the high-speed train of AI to actually run. The output gap is a temporary, necessary phase. It is the silence before the orchestra starts playing. The potential energy is building up in the form of capability, but it cannot convert into kinetic economic energy until the friction is reduced. The “stalled” revolution is simply a revolution that is currently under construction. We are not witnessing the ceiling of AI intelligence; we are witnessing the floor of AI adoption. The technology has done its part; now, the organizations must do theirs. The future isn’t late—it’s just waiting for legal to sign off.

Dwarkesh is WRONG about the "Output Gap" The narrative around Artificial Intelligence has shifted perceptibly in late 2025. After years of exponential hype, a sense of disillusionment has begun to settle over the industry. Commentators and analysts, most notably podcaster and writer Dwarkesh Patel, have recently highlighted what is being called the “Output Gap.” This is the uncomfortable discrepancy between our models’ skyrocketing performance on benchmarks and the relatively stagnant growth in macroeconomic productivity. We have reached “superhuman” capability on tests, yet global GDP hasn’t skyrocketed, and the promised revolution feels curiously delayed. This frustration stems from a fundamental misunderstanding of where we are in the technology cycle. The industry is currently fixated on “Day 0” capabilities—the raw intelligence of the models, the scaling laws, and the saturation of academic benchmarks. However, the bottleneck has shifted. We are no longer limited by the intelligence of the model, but by the inertia of the enterprise. The “Output Gap” is not a failure of technology; it is a lag in organizational digestion. We have invented a powerful jet engine, but we are essentially frustrated that it hasn’t revolutionized travel before we’ve even built the airframe to mount it on. The primary fallacy driving this disappointment is the expectation of the “drop-in remote worker.” Many observers equate Artificial General Intelligence (AGI) with a digital human that can be onboarded, culturally assimilated, and left to run autonomously with minimal supervision. Because current agents cannot seamlessly replace a human employee in this one-to-one fashion, the conclusion is often that the technology has stalled. This view misses the forest for the trees. Disruptive technologies rarely act as direct replacements; instead, they require a complete restructuring of how work is done. In reality, the barrier to adoption is what IT professionals call “Day 2 Operations.” Day 0 is the exciting launch; Day 2 is the boring, messy reality of governance, security, and maintenance. For an enterprise to deploy an autonomous agent, it isn’t enough for the model to be smart. The organization must solve for Role-Based Access Control (RBAC), SOC 2 compliance, liability frameworks, and data sovereignty. Right now, most organizations lack the infrastructure to handle “non-human” identities that have access to sensitive corporate data. Consider the security implications. A human employee has physical limitations and a single identity. An AI agent is an “always-on” entity that, if given improper permissions, could theoretically read every email in a company server to “optimize workflow.” Security teams (CISOs) are rightly terrified of this prospect. Until we develop granular access controls specifically designed for agents—effectively an “RBAC for AI”—the widespread deployment of autonomous agents will remain blocked by the “Departments of No”: Legal, HR, and Security. This operational friction explains why we are seeing a massive divergence between individual and enterprise adoption. Individually, adoption is rampant; we are in the “Shadow IT” era of AI, where employees secretly use tools like ChatGPT to boost their personal productivity. However, at the organizational level, adoption is glacial because the institution’s primary mandate is risk management, not speed. The C-suite is asking questions about ROI, liability, and data leakage that the current software ecosystem cannot yet answer satisfactorily. History offers a comforting precedent for this timeline. We are currently effectively in the “2002 era” of virtualization. In the early 2000s, virtualization technology (like VMware) was technically viable, but it took nearly a decade to become the default enterprise standard. It required years of maturing management software, security protocols, and cultural shifts before “The Cloud” became a reality. AI is undergoing the same digestion period. The technology is ready, but the enterprise “rails” required to run it are still being laid. This leads us to the “Mechanical Horse” fallacy. When the automobile was invented, people didn’t need a mechanical horse that walked on four legs; they needed a car. But the car required asphalt roads, not dirt paths. Similarly, we are currently trying to force AI into human-shaped workflows (”jobs”) rather than rebuilding the workflows to suit the AI. We are waiting for the mechanical horse, failing to realize that the true disruption comes from unbundling the work entirely. A “job” is essentially a bundle of tasks, context, and responsibilities aggregated for a single human. AI does not replace jobs; it unbundles tasks. The transition we are navigating involves breaking down these bundles and determining which specific value streams can be automated by agents. This requires task decomposition and new management frameworks—a “Council of AIs” approach—rather than a single omnipotent bot. This restructuring is an organizational physics problem, not a computer science problem. The disconnect is further exacerbated by the differing timelines of developers versus executives. Developers and researchers live in a world of root access and rapid iteration, often blinded to the molasses-like speed of corporate change management. A CFO, conversely, thinks in fiscal quarters and compliance audits. They require proven ROI and standardized best practices before signing off on widespread automation. Currently, there are no industry standards for deploying autonomous agents, which halts most conversations at the boardroom door. Therefore, the “stalling” progress is an illusion caused by looking at the wrong metrics. If you look at model cards and benchmarks, progress is linear and fast. If you look at widespread economic integration, the curve is flat. This S-curve of adoption is always significantly behind the S-curve of capability. We are in the flat part of the adoption curve, characterized by high hype, high friction, and frantic infrastructure building behind the scenes. The next three to five years will likely be dominated not by flashy new model capabilities, but by the “boring” work of integration. We will see the rise of startups and consultancies dedicated solely to “AI Governance,” “Agent Identity Management,” and “Context Orchestration.” These are the boring rails that will eventually allow the high-speed train of AI to actually run. The output gap is a temporary, necessary phase. It is the silence before the orchestra starts playing. The potential energy is building up in the form of capability, but it cannot convert into kinetic economic energy until the friction is reduced. The “stalled” revolution is simply a revolution that is currently under construction. We are not witnessing the ceiling of AI intelligence; we are witnessing the floor of AI adoption. The technology has done its part; now, the organizations must do theirs. The future isn’t late—it’s just waiting for legal to sign off.

David Shapiro (L/0)

12,656 просмотров • 9 месяцев назад

Here's why $NEAR is a no-brainer in 2025 👇 Everybody loves NEAR Protocol and there is a reason for that (or many). Near is well-positioned to be one of the leading blockchain ecosystems this year. Let’s explore the “whys”. TIMESTAMPS Quick Bio – 00:00:15 Inflation Reduction Proposal – 00:00:43 Technically Speaking – 00:02:40 Near Intents – 00:03:37 Chain Signatures and AI – 00:04:39 Decentralization and DeFi – 00:05:59 I have my Near account since March 2023, but it has been inactive for a while, as I was focused on other stuff. However, the recent inflation halving proposal by HOT DAO (HOT Protocol 🔥) and LiNEAR (LiNEAR Protocol) brought my eyes back to the project and I really like what I’m seeing. So, here’s my first point. If this proposal passes, NEAR could lead the way in what appears to be a market trend of improving the tokenomics, as more and more experts realize holders have been overpaying for these networks' security, with a too high supply inflation. Solana tried something similar, but the proposal was rejected. In my opinion, validators voting favorably to that show a commitment to the chain for the long term. On the other hand, voting against it signals a short-term vision focused on milking the emissions as much as possible, at the ecosystem’s expense. The voting currently goes with 28% “YEA” votes, needing 66.76% to pass. Most of the validators who already cast their votes went with the yes. 2pilot, avb, openshards, qbit, sicmundus, fox, and intear are, so far, the only seven who voted “NAY”. This proposal has the vocal support of most influential figures in the Near ecosystem, including the Near Foundation (NEAR Foundation), led by Illia (root.near) (🇺🇦, ⋈), which makes me believe it will pass and show the power of the halving in getting the market’s attention and presenting a huge investment asymmetry for the native token right now. Is this everything I like about NEAR? Definitely not. This is just what got me looking at it again, just to discover a (very much) thriving ecosystem, full of interesting things happening at the same time. I’ll mention a few, but there is (much) more. Technically speaking, Near is a high-performance blockchain, with really low fees and one of the fastest finalities, with 600ms block time and approximately 1.8s finality. It also has my favorite architecture for internet-scale scalability, using sharding, while keeping a high decentralization standard. As a learning programmer, Near also has one of the best dev experiences (in my limited opinion). The documentation is clear, has a logical journey, presenting from the basic anatomy in details to more complex SDKs and tools. I’m also in love with the near-cli-rs. A command line interface program written in Rust for seamless interaction with the Near blockchain. Allowing wallet creation, chain query, sending transactions, staking, smart contract calls, and more. Near Intents. This was the second thing to get my attention, while studying the project again, and it sets a whole new standard for blockchain interactions, especially cross-chain. Basically, users can declare an intention (for example, swap Ethereum-USDT to Bitcoin) and a network of solvers, running on Near, will find the best path to accomplish this task. We recently saw an impressive 465k-worth swap happening in exactly this example, paying 0.55% of trading fees to thorswap.near and swapkit.near. According to a Dune Dashboard, the protocol accumulates nearly $400 million in volume since its launch not long ago, in November 2024. *obs.: half this volume was achieved in the last month. Massive! Near Intents is possible due to two other very interesting things: (i) Chain abstraction, and (ii) a solid AI infrastructure. Chain abstraction (via Chain Signatures) is a powerful interoperability feature, allowing Near to friendly connect different blockchains as if they were part of a single network. Users and devs benefit from wallet, address, fees, and cross-chain bridges abstractions - not even noticing they are interacting with multiple chains. One wallet that powers everything. Powered by Near. On AI, Near is just built differently. Not for the hype, but for the solution. The team has been looking for AI solutions much before the ChatGPT fever. Actually, they started as an AI company, pivoting to blockchain later. So, being one of the most promising networks for the growing AI economy was just the natural path to follow. There is an extensive and super complete research piece on that topic, recently published by Reflexivity Research (Reflexivity Research) on July 1st. It presents Near as an AI-optimized blockchain, covering AITP, Shade Agents, x402, Near Intents, and more. Definitely worth the reading. Wrapping up this content with one more aspect that really matters to me is how Near remains truthful to decentralization, data ownership, censorship-resistance and open-source primitives that have been increasingly abandoned by other key players. A simple example of that is how the Near Foundation decided to deprecate its public APIs, encouraging the surge of a more decentralized and competitive market of SaaS projects, with a highlight to Lava Network, that recently appeared in my timeline talking about that. DeFi is also huge on Near, leveraging all the previous properties I mentioned, creating a truly decentralized liquidity pool via Rhea Finance, connected with other chains like BTC, Ethereum, ZCash, and more. All that contributes to Near having the second-largest monthly active addresses, with nearly 50 million, only losing to Solana’s nearly 90 million. In the meantime, NEAR, the token, is not even at the 30rd position by market cap. Crazy stuff. To (finally) wrap it up, I also want to mention Near’s consensus decentralization. While having a low node-count, the network has a Nakamoto Coefficient of 11, which is not bad at all. Surely, there is still room for improvement, which is possible as becoming a validator is accessible staking and hardware-wise. If you liked this content, make sure to click the like bottom and share it around. Follow me on X or subscribe to my YouTube channel, both at vinibarbosabr. See ya!

Here's why $NEAR is a no-brainer in 2025 👇 Everybody loves NEAR Protocol and there is a reason for that (or many). Near is well-positioned to be one of the leading blockchain ecosystems this year. Let’s explore the “whys”. TIMESTAMPS Quick Bio – 00:00:15 Inflation Reduction Proposal – 00:00:43 Technically Speaking – 00:02:40 Near Intents – 00:03:37 Chain Signatures and AI – 00:04:39 Decentralization and DeFi – 00:05:59 I have my Near account since March 2023, but it has been inactive for a while, as I was focused on other stuff. However, the recent inflation halving proposal by HOT DAO (HOT Protocol 🔥) and LiNEAR (LiNEAR Protocol) brought my eyes back to the project and I really like what I’m seeing. So, here’s my first point. If this proposal passes, NEAR could lead the way in what appears to be a market trend of improving the tokenomics, as more and more experts realize holders have been overpaying for these networks' security, with a too high supply inflation. Solana tried something similar, but the proposal was rejected. In my opinion, validators voting favorably to that show a commitment to the chain for the long term. On the other hand, voting against it signals a short-term vision focused on milking the emissions as much as possible, at the ecosystem’s expense. The voting currently goes with 28% “YEA” votes, needing 66.76% to pass. Most of the validators who already cast their votes went with the yes. 2pilot, avb, openshards, qbit, sicmundus, fox, and intear are, so far, the only seven who voted “NAY”. This proposal has the vocal support of most influential figures in the Near ecosystem, including the Near Foundation (NEAR Foundation), led by Illia (root.near) (🇺🇦, ⋈), which makes me believe it will pass and show the power of the halving in getting the market’s attention and presenting a huge investment asymmetry for the native token right now. Is this everything I like about NEAR? Definitely not. This is just what got me looking at it again, just to discover a (very much) thriving ecosystem, full of interesting things happening at the same time. I’ll mention a few, but there is (much) more. Technically speaking, Near is a high-performance blockchain, with really low fees and one of the fastest finalities, with 600ms block time and approximately 1.8s finality. It also has my favorite architecture for internet-scale scalability, using sharding, while keeping a high decentralization standard. As a learning programmer, Near also has one of the best dev experiences (in my limited opinion). The documentation is clear, has a logical journey, presenting from the basic anatomy in details to more complex SDKs and tools. I’m also in love with the near-cli-rs. A command line interface program written in Rust for seamless interaction with the Near blockchain. Allowing wallet creation, chain query, sending transactions, staking, smart contract calls, and more. Near Intents. This was the second thing to get my attention, while studying the project again, and it sets a whole new standard for blockchain interactions, especially cross-chain. Basically, users can declare an intention (for example, swap Ethereum-USDT to Bitcoin) and a network of solvers, running on Near, will find the best path to accomplish this task. We recently saw an impressive 465k-worth swap happening in exactly this example, paying 0.55% of trading fees to thorswap.near and swapkit.near. According to a Dune Dashboard, the protocol accumulates nearly $400 million in volume since its launch not long ago, in November 2024. *obs.: half this volume was achieved in the last month. Massive! Near Intents is possible due to two other very interesting things: (i) Chain abstraction, and (ii) a solid AI infrastructure. Chain abstraction (via Chain Signatures) is a powerful interoperability feature, allowing Near to friendly connect different blockchains as if they were part of a single network. Users and devs benefit from wallet, address, fees, and cross-chain bridges abstractions - not even noticing they are interacting with multiple chains. One wallet that powers everything. Powered by Near. On AI, Near is just built differently. Not for the hype, but for the solution. The team has been looking for AI solutions much before the ChatGPT fever. Actually, they started as an AI company, pivoting to blockchain later. So, being one of the most promising networks for the growing AI economy was just the natural path to follow. There is an extensive and super complete research piece on that topic, recently published by Reflexivity Research (Reflexivity Research) on July 1st. It presents Near as an AI-optimized blockchain, covering AITP, Shade Agents, x402, Near Intents, and more. Definitely worth the reading. Wrapping up this content with one more aspect that really matters to me is how Near remains truthful to decentralization, data ownership, censorship-resistance and open-source primitives that have been increasingly abandoned by other key players. A simple example of that is how the Near Foundation decided to deprecate its public APIs, encouraging the surge of a more decentralized and competitive market of SaaS projects, with a highlight to Lava Network, that recently appeared in my timeline talking about that. DeFi is also huge on Near, leveraging all the previous properties I mentioned, creating a truly decentralized liquidity pool via Rhea Finance, connected with other chains like BTC, Ethereum, ZCash, and more. All that contributes to Near having the second-largest monthly active addresses, with nearly 50 million, only losing to Solana’s nearly 90 million. In the meantime, NEAR, the token, is not even at the 30rd position by market cap. Crazy stuff. To (finally) wrap it up, I also want to mention Near’s consensus decentralization. While having a low node-count, the network has a Nakamoto Coefficient of 11, which is not bad at all. Surely, there is still room for improvement, which is possible as becoming a validator is accessible staking and hardware-wise. If you liked this content, make sure to click the like bottom and share it around. Follow me on X or subscribe to my YouTube channel, both at vinibarbosabr. See ya!

Vini B |「 thecoding 」

40,183 просмотров • 1 год назад