gAxis guys!! Drawer tasks in @axisrobotics can feel weirdly annoying at first .. But once you figure out the right order, they become much easier. Here’s the trick I use 👇 Don’t start with the top drawer. Open the lower one first. Also, double-tap objects and drawers whenever you can it saves a lot of unnecessary movement. One small control makes a big difference too: Hold 'X' to tilt the claw upward without moving the entire arm. This makes grabbing drawer handles much easier. My usual flow is: ----------------- → Open the lower drawer → Go to the upper drawer → Double-tap the object you need → Grab it properly → Double-tap the drawer to move back quickly → Put the object inside the upper drawer → Close it Once you get used to this sequence, the task becomes much smoother. And if you haven’t joined the rewards preprogram yet, you’re still early: hub.axisrobotics.ai/login?in… Bookmark this for your next run, and let me know if you’re stuck anywhere. @axisrobotics

The Service Layer for Physical AI: ⚫ Where Teleoperation Becomes Training Data Building a capable robot is not only about the robot itself. A big part of Physical AI is what happens behind the scenes how robots collect real-world experience, how humans guide them, and how that experience becomes useful training data. This is where teleoperation becomes really important. Instead of asking a robot to figure out every task on its own, a human operator can control or guide the robot through real-world actions. The robot can then capture those movements, decisions, camera views, and interactions as data. But simply collecting teleoperation data is not enough. The data needs to be organized, checked, validated, and prepared before it can actually help train better models. That is where a service layer for Physical AI can become valuable. Think of the process like this: ....................................................... Human → Teleoperation → Robot Data → Validation → Training → Better Robot Every step adds something important. Teleoperation provides real-world demonstrations. Validation helps separate useful demonstrations from poor ones. And high-quality data can then be used to improve the models that power future robots. This creates a continuous loop where humans help robots learn, robots generate more data, and that data helps build better Physical AI. For me, this is one of the interesting parts of the Physical AI stack. The future may not just be about building smarter robots. It may also be about building the infrastructure around them that turns human experience into reliable machine intelligence. @PrismaXai

🔴 How Optimum Uses RLNC to Rethink Blockchain Networking ... Most people think blockchain scaling is only about making transactions execute faster. But there’s another challenge happening behind the scenes: how quickly information travels between nodes. That’s the problem Optimum is working on. Instead of repeatedly sending the same data across the network, Optimum uses Random Linear Network Coding (RLNC). Data is encoded into different combinations, so nodes can rebuild the original information once they receive enough useful packets even if some packets are lost. Why does that matter? * Faster propagation of blocks, blobs, and transactions * Less duplicate network traffic * Better resilience to packet loss * More efficient use of bandwidth * Stronger networking for high-performance blockchains What I like most is that Optimum focuses on the networking layer, not replacing a blockchain’s consensus or execution. The future of scaling isn’t only about processing data faster it’s about moving data smarter. @get_optimum

gMum guys!!! Imagine a busy highway where every car is carrying the exact same package. More cars fill the road, traffic gets heavier, and everyone takes longer to reach their destination. Blockchain networks can face a similar problem. When the same data is sent repeatedly between nodes, it creates extra bandwidth usage and unnecessary delays. That’s why Optimum is taking a different approach. Using RLNC-based networking, it helps blocks, blobs, and transactions move across the network more efficiently. Instead of relying on repeated transmissions, the network is designed to recover data in a smarter way. The idea is simple: ------------------- Less redundant traffic. Faster data delivery. A stronger blockchain network. @get_optimum

Just tried @FUDmarkets on Solana and opened my first BTC LONG. I deposited $1.10 USDC and took a small BTC position with $0.5535, ending up with 0.97 shares. The whole interface was pretty easy to understand. I could quickly check my open position, exposure and P&L from the profile page, and seeing the position confirmed on Solana was nice. I also liked that the position details are kept simple without making the screen feel too complicated. My first trade is currently showing around -$0.0491 P&L, so now I’m curious to see how the position plays out .. Overall, the experience felt straightforward, and I’d like to try a few more markets on FUD Markets. @GenLayer @FUDmarkets

gPrisma guys!! No Robot Is Safe And That’s Exactly Why AI Safety Matters !!?? When we talk about AI safety, we usually think about chatbots giving wrong answers, harmful instructions, or unreliable information. But physical AI is a different story. A language model making a mistake might give you a bad answer. A robot making a mistake can actually move, break something, damage equipment, or even hurt someone. That means physical AI needs a much higher level of reliability. One small movement error can sometimes become a much bigger problem. A robot arm pushing in the wrong direction, moving too fast, or performing the wrong action can cause real-world damage. So safety for robots cannot depend only on the AI model. Things like soft stops, safety controllers, geofencing, physical limits, and careful deployment are also important. These systems can prevent a robot from going somewhere or doing something dangerous, even if the model makes a mistake. There is also another important part: the training data. If we don't want robots to learn dangerous behaviors, we should be careful about what goes into their training data in the first place. The goal isn't to make robots perfect. The goal is to build enough layers of validation, training, and physical safety that one mistake doesn't turn into a real-world disaster. As robots become more capable and autonomous, safety can't be an afterthought. It has to be part of the system from the very beginning. @PrismaXai

Guys, Missions #14 and #15 are officially live on ZenO ... This time, you’ll be working with everyday kitchen tasks: Mission #14 - Refrigerator Organizing Take items out of the fridge, sort them, move things around, and organize the food and containers properly. Mission #15 Food Preparation Get your ingredients ready by washing, cutting, peeling, measuring, and doing the basic prep before cooking. Before you start recording, make sure to check the updated mission guide. The Steps, Required, and Do Not sections have all been revised, so following the latest instructions is important. For both missions, remember: • Phone must be head/face mounted • Record in landscape • Keep both hands visible throughout • First-person POV only • Use 0.5x–0.7x wide angle Once your recording is ready, upload it through ZenO Core. Every small action you capture can become useful data for training the next generation of Physical AI. Record carefully. Follow the guidelines. Contribute to smarter robots. @ZenO4AI

gMum guys!! New week, new role upgrades in the Optimum community! Big congratulations to everyone who earned the Refined role this week. These upgrades are a nice reminder that consistent effort really matters. Being active, helping others, sharing ideas, and contributing to the community can all make a difference over time. It’s not about showing up once and disappearing. It’s about staying involved and growing together. Congrats to everyone on the list! Keep building, keep learning, and keep pushing forward. @get_optimum

Good afternoon guys!! When we talk about blockchain scalability, the first thing that usually comes to mind is execution. But there’s another part that deserves more attention: how fast can data actually move between nodes? A block can be produced quickly, but if it takes too long to reach validators across the network, that can still create delays and unnecessary network traffic. This is the problem Optimum is focusing on. Its approach is to improve the networking and data propagation layer, helping blocks, blobs, and transactions move more efficiently without requiring the whole blockchain to be redesigned. I find this direction interesting because scaling isn’t only about doing more work on-chain. Sometimes, the network simply needs to move information better. @get_optimum

The Bunii world is open Free mint on Robinhood Closes in 48hrs

Here we goo 🧪 We’ve teamed up with RisePunks and secured 10 WL spots for our community. RisePunks is bringing 2,000 Punks to RISE, and if you’ve been around the RISE ecosystem, this one might already be on your radar. • Supply: 2,000 • Chain: RISE • Mint: September 24 • WL Price: TBD, but well below 0.009 ETH • 10 WL spots for AlphaBroz To enter: • Follow @RisePunks & @alphabrozdao • Like + RT this post • Drop your EVM wallet below Remember, you’ll need to bridge ETH to RISE to mint. 10 spots up for grabs. Good luck BROZ 🤝

i’ve been [ ZDACTED ]. yours is still ████████. @zdacted zdacted.xyz/r/ZD-ZQCKKPG7SJ7…

SEPTEMBER FREE MINT GIVEAWAY IS COMING - Meet H00D KEY 999 FREE MINTS September 25, 2026 Dropping on @opensea 🔸 How to Enter : ⚪ Follow - @Duckling_nft & @isturaf ⚪ Repost ⚪ Drop your EVM address in the comments 👇 Get ready for September 25.

whitelist applications: closed. And, 21,996 applied. Obviously it's so huge.. My all friends got approved but me still pending 😞.. Wen me intern @zaddrnet

Hey, I just joined the ZODD NFT whitelist 🐸 — ZEC to zillions, privacy is normal 💚🔒 #Zcash $ZEC

As planned, we’ve officially decided to launch our NFT marketplace featuring the ZODD Relics. Step into the Zoddverse and apply for the WHITELIST. zodd.fun/nft/whitelist Don't miss out on this alpha opportunity! Drop date coming soon. Zoldiers! 🛡️🚀 #Zcash #ZEC #NFT $ZEC

10k ghosts. Zero faces. Proven on-chain. Identity off. @zkghosts_

New operation loading guys!! Let's ready

I applied two days ago and now still pending 😞😞. Wen approved me @zaddrnet

Sim Teleop: A New Way to Teach Robots ------------- One thing I’ve been thinking about with robotics is this: How do you teach a robot to do something it has never done before? You can give it code and instructions, but physical tasks are often much more complicated than they look. A human can look at an object, understand its position, move their hand, adjust their movement, and complete the task almost naturally. For a robot, every one of those steps can be a challenge. That’s why simulation and teleoperation are so interesting. With Zen-O’s Sim Teleop, humans can interact with a robot inside a simulated environment and guide its actions. Instead of only telling a robot what to do, we can show it how the task is performed. A simple interaction can create a useful demonstration: Human action → Robot movement → Recorded experience → Training data And doing this in a simulated environment has some clear advantages. You can test different movements, repeat tasks, change the environment, and experiment without needing a physical robot for every single attempt. That means more room to learn and improve. What I like about this idea is that it connects human intelligence with machine learning. Humans are good at understanding situations and adapting to changes. Robots are good at repeating tasks and processing huge amounts of information. When you combine the two, the learning process becomes much more interesting. For Physical AI to grow, robots need ways to learn from real human behavior and different situations. Simulation + Teleoperation + Quality Data = a powerful path toward smarter robots. Zen-O is working on an interesting piece of that puzzle. @ZenO4AI