Vladyslav Semenchenko

I’ve spent the last two months with the @Meta @ray_ban AI glasses. It’s a genuinely great product. Here’s what I’m still missing: 1) instant answers. I understand the compute limitations of on-device AI, but even flash responses to simple questions or real-time translation would dramatically improve the experience. Current speed around 3-5 seconds for simple questions or translation is good but not great. 2) knowledge aggregation. I want to effortlessly save my ideas, things I see around me, personas, and other context. The current workaround — using a WhatsApp bot that talks to my computer, stores, and categorizes everything - actually works surprisingly well (including asking about past things). But it’s not an out-of-the-box experience. Nailing this natively would be a game-changer for productive users. 3. “Always-on” capturing (video, audio, or both). The rumored feature would be fantastic (Limitless acquisition to the rescue!). Overall, the experience is astonishingly good. The moment you’re riding a bike, spot an unfamiliar building, and can just say “Hey Meta, what’s this?” without even slowing down - and instantly get an answer - feels like pure magic. Or saying “Hey Meta, take a photo and translate this” without ever touching your phone… it genuinely feels like the future. Devices like these could eventually replace phones for many people.

I can’t say enough good things about John Carmack @ID_AA_Carmack and his Keen Technologies. But now Khurram Javed @kjaved_ and I have broken away to start our own startup and pursue a slightly different path toward understanding intelligence. Like Keen (and like Ineffable) we at Oak Lab @oaklab_ai believe in reinforcement learning and that intelligence is created and maintained from run-time experience. But we think current deep learning methods are weak and inefficient, and need not more tweaks, but fundamentally new ideas and a thorough reworking before they can provide a solid foundation for achieving the more ambitious goals of AI.

Apple's M7 Ultra chip coming in 2029 is rumored to support 1.5TB of RAM. This would make the processor much more capable for on-device AI.

What is planned in terms of iOS tech stack - iOS 27 Siri support to interact with the App natively through voice/chat.

Got Liquid Glass support in my app. Alongside with: dark theme support, complete set of accessible features (reduced motion, voice over, dynamic type), complete UI overhaul (100% SwiftUI and modern concurrency), widgets, quick actions, spotlight actions and many other small changes!

I thought that’s a joke but no - it’s real! 😂😂😂

@FloWritesCode @cursor_ai Nicely done! Where did you get image assets? Generated (where), downloaded, created?

I've been building a little mobile RPG game with @cursor_ai It's still in its very early days, but here's the concept: - Infinite progression - Short sessions (2-5min, multiple times a day) - Procedural dungeons - Mostly PvE - Stats > Skill It's all made in SwiftUI - even the dungeon you can see in the first screenshot. Rendering is based on Canvas and TimelineView and it's running super smooth!

@sxpstudio Because even with unlimited tokens there are a lot of humans behind AI (aka operators) :) that’s why they are still hunting people and not able to delegate all tasks to own AI solutions (which is wise of course)

If you're a company with unlimited tokens, I honestly don't understand why you'd be going with electron and that sort of nonsense. This is the one time where the sempiternal debate of "I have to write my UI twice" is actually doable.

@rudrank @mopperson Glad you found out! Yes, that's AFM 3 Core Advanced which has 8K context. When you use `SystemLanguageModel` you get AFM 3 Core Advanced on these devices, and AFM 3 Core on older devices.

@rohanpaul_ai Very close to human behavior! :)

Humanoid in Shenzhen, China. Real-time stability management is among the toughest problems in developing reliable legged robots outdoors.

@did0f Kindle for sure. I am using an old (2020) Oasis but it works well (IMO it has the best ergonomics of any Kindle devices). Tried new Colorsoft but it wasn’t a game changer for me. Grab any new Kindle model and you will be happy with it!

I'm getting back into the habit of reading. However, paper books are too bulky; I need an eReader. What is your favorite? It would be nice if it could be used to highlight and bookmark unknown words or translate them. I’m open to change my mind, by now I am looking at Kindle Paperwhite 16 GB, 2024 model.

I spent around 2k on a Sony home theater audio system (7.1.2 or 9.1.2 - I don’t remember exactly) and I’ve never regretted it. Every time I listen through it, tears come to my eyes. What an amazing way to actually feel the music! By the way, it’s hard for me to go back to AirPods or even the XM6 now:) So I mostly listen to podcasts on headphones these days:)

discovered dolby atmos remasters recently and I regret to inform you that music apparently has been in black and white this whole time. even on my budget, sub $1k setup (2x sonos era 300s) albums i've heard 500 times suddenly grew a third dimension, almost making me cry. WHY IS NOBODY TALKING ABOUT THIS. switched from spotify to apple music purely for this and I'd do it again in front of daniel ek's family. its insanely good - do yourself a favor and go hear it.

In the future, you will purchase knowledge models, not robots. Robots will be too cheap, but models save you a lot of time. Just imagine, you are unboxing your new TS3 robot. You have an option to start from scratch and train/teach it. It will be like a child who knows only something too primitive. Or you have an option to purchase a knowledge model of a 3 yo person, 5 yo person, 9, 14, 18, 23, and so on. So basically, you will not purchase a robot but a persona, a knowledge. For companies, it will be cheaper to buy an already mature person/model, put it in production, and give it tasks. For enthusiasts, they can start from scratch and experiment.

Two insights from 2140: 1) elastic displays have bad UX, it’s hard to long press on them in stressful situations 2) #AMZN @amazon is still working!

$MU @MicronTech has had an amazing run. I remember when I bought it for around $100. Now the price is $1,100! I also remember when people said that 10x returns could only be made in the crypto market. In my opinion, every market matters. It’s better to keep an open mind and watch for opportunities everywhere!

I see a lot of brokers spamming notifications about SpaceX IPO. It simply means that they will short, isn’t it? :)

Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use. Its capabilities exceed those of any model we’ve ever made generally available.

Worth to watch! youtu.be/K5LAFEjTlBA?is…

Today we are all computer/agent operators. Do AGI will need operators..? 🤔