Don Vito

“Please remain seated.” @NeoSoulAI @MollyApriltian The carousel horse took that personally. 🎠👻 Meet AntBoy, the ringmaster of a Halloween carnival he is absolutely unqualified to run. I dressed him in a purple tailcoat, gold epaulettes, white gloves, and a feathered hat with an eye of its own. A grand entrance deserves a grand outfit. 💜✨ Unfortunately, one enchanted horse escaped the carousel before his opening speech. Now he’s carrying the problem around, hoping nobody notices it has teeth. The tickets said “an unforgettable night.” Technically, he delivered. 🎃😂 My entry for the Halloween AntBoy COS Challenge! #NeoSoulAntBoyHalloween

How @PrismaXai Is Creating Millions of New Jobs in the Physical AI Era 1/ Physical AI needs people more than people realize. Robots don’t learn from code alone. They learn from humans showing them how to do real things: pick up objects, fold clothes, recover from mistakes, work in actual rooms. PrismaX turns that into paid work. 2/ Most companies can’t build this themselves. Finding skilled people, training them, checking quality, and paying them fairly is hard. That’s why so much robotics data is low quality or just doesn’t exist at the scale models need. PrismaX built the layer that handles it. 3/ There are a few simple ways to get paid. Operate real robot arms from home, on your own schedule. You guide the robot through tasks. Valid sessions get paid. Record everyday work with a normal camera or headset. No robot needed. Clips that meet the standard earn money. Review and score other people’s demonstrations. Good reviewers help protect quality and get rewarded for it. 4/ The barrier is low on purpose. You don’t need a special degree. You need hands, a bit of care, and the willingness to do clean work. PrismaX handles the standards, the matching to real demand, and the payments. Some people own robots and put them to work. Others just use a camera. Both count. 5/ Quality is what keeps the jobs real. Not every clip gets paid. Only the ones that are actually useful for training. Smooth movement, clear intent, real recovery when something goes wrong. That standard is what makes companies keep buying the data instead of trying to collect it themselves. 6/ This is how it scales. The more robots get deployed in the real world, the more they need fresh demonstrations of new tasks and new environments. Every new capability creates more demand for the next round of human teaching. PrismaX is building the network that can supply that demand without every robotics company having to hire its own team. 7/ Most people hear “robots” and think jobs disappear. In the data layer, the opposite happens. Humans become the teachers. The better the system gets at turning human skill into training data, the more people can participate from anywhere. Operators, recorders, and reviewers all sit inside the loop instead of outside it. 8/ The work is already happening. People are operating robots remotely. Others are recording ordinary tasks. Reviewers are scoring quality so only the good data moves forward. Demand from teams building physical AI is what drives the whole thing. It’s early, but the shape is clear: a new category of work that grows as the technology improves. 9/ If you want to teach machines instead of just watching them, this is one of the few places doing it properly. Clear standards. Real demand. Money flowing back to the people who create the data. The Physical AI era needs millions of people who can show robots how the world actually works. PrismaX is building the system that employs them.

Official collab: @NeoSoulAI × Renaiss On one side, @renaissxyz is creating a proper marketplace for real world collectibles bringing physical items onchain with proper ownership, grading verification, and redemption rails. @Hansmahanakon On the other side, NeoSoul is building NeoTrade so everyday users can spin up and run their own AI trading assistants. @MollyApriltian Together, we’re looking at how AI agents can actually help people navigate the collectibles space more intelligently from discovery to timing decisions. More to come on the product side and specifically across BNB Chain.

the day i'll never forget in crypto MY NEOSOUL REWARD, MY NEW BIKE. 🚴 I just bought myself a new bicycle with the airdrop rewards I received from @NeoSoulAI. It's not a supercar or anything luxurious. But it's one of the most worthwhile purchases I've made since entering crypto. During Season 1, I decided to go all in with my time and effort on NeoSoul. Instead of chasing dozens of projects, I focused on learning about AI Agents, creating content, participating in community activities, and following the project's development closely. Some people asked me: "What if you spend all that time on one project and end up getting no airdrop?" My answer was simple: "I've made my choice. Now I'll give it my best." Of course, I expected a reward. When you put real time and effort into something, you naturally hope to see results. And NeoSoul didn't disappoint me. The day I received my airdrop, I knew my decision had paid off. Instead of reinvesting everything into crypto, I used part of the reward to buy a new bicycle, something I'd wanted for a long time to improve my fitness. Working in Web3 means spending way too much time in front of a screen. Some days, my entire routine revolves around X, Discord, and writing content. Now I have another reason to get outside every morning, stay active, and start the day with a clear mind. This bike is also my way of rewarding myself after a season of serious effort. What I respect about @NeoSoulAI is that the project doesn't just talk about AI Agents. It also creates opportunities for community members to participate, learn, contribute, and be rewarded for their efforts. For me, Season 1 was worth it. I gained knowledge, built new connections, and received a reward I could actually put to use. Thank you, NeoSoul, for recognizing my contributions. Season 1 helped me buy this bicycle. In Season 2, I'll keep showing up, working hard, and aiming to contribute even more. I don't need to say much more about the results. The bike is sitting at my place. And tomorrow morning, I'll be out riding it again. 🚴💪

Why @PrismaXai is quietly one of the most important infrastructure plays in physical AI right now Most physical AI projects still hit the same wall. You can train a big model. You can run impressive demos. But when the robot has to deal with real friction, real lighting, real object variation, and real human environments, performance drops. The missing piece has always been high quality demonstrations of how the physical world actually behaves. PrismaX is building the layer that supplies that. What makes their approach different is that they treat data as an active system, not a static dataset. Teleoperation sessions and egocentric recordings are collected against specific tasks, scenes, and robot types. The data isn’t generic video. It’s structured so it can actually be used to train vision language action models. That distinction matters when you’re trying to get a policy that transfers instead of just looking good in a paper. Egocentric data is underrated here. A person puts on a headset or holds a camera and records ordinary hands on work. The resulting first person video captures the exact visual and motion signals a robot needs to learn the same task. No robot required on the collection side. The system still ends up with usable demonstrations that meet a clear quality standard. Teleoperation closes the other half. Skilled operators control real hardware on real tasks. The sessions produce the kind of contact rich, multi step demonstrations that synthetic data and lab collected footage usually miss. Because the operators are working against actual demand, the data stays grounded in what builders actually need instead of drifting into academic tasks. The interesting technical piece is the feedback loop. Collection isn’t one way. Teams train on the data, evaluate where the model fails, and that signal feeds back into what gets collected next. Over time the system gets better at knowing which kinds of demonstrations actually improve real-world performance. That’s how you move past “more data” and toward useful data. They’ve also made the data practical to work with. Searchable by task and scene. Downloadable. Available at different quality tiers. Custom collection when off-the-shelf doesn’t match what you need. For teams that previously had to stand up their own operators, cameras, review processes, and storage just to get a few hundred hours of usable footage, this removes a large amount of undifferentiated engineering work. Physical AI models are hungry for diverse, proven examples of contact, failure recovery, and multi-step reasoning in the real world. PrismaX is focused on producing exactly that not more simulation, not more web-scale video, but reviewed demonstrations that come from real human skill applied to physical tasks. That focus is why the data tends to be more usable when you actually fine-tune. The broader idea is a shared service layer. Instead of every robotics team reinventing data collection, quality review, and distribution, there’s a common substrate that supplies structured, high-intent demonstrations. When that layer exists, more teams can spend time on model architecture, deployment, and hardware instead of rebuilding the same data pipeline. None of this is flashy in the way humanoid demos are flashy. It’s the unglamorous work of turning human skill into clean training signal, then making that signal available so others can train on it. But that work is what determines how fast physical AI moves from controlled demos to robots that reliably do useful things in messy environments. PrismaX is solving the data problem the way infrastructure should: with clear standards, real-world collection methods, and a path for the data to improve over time. If you’re building anything that has to interact with the physical world, this is the kind of technology that quietly makes everything else possible. Worth paying attention to.

WEB3 GAVE ME MORE THAN I EXPECTED ❤️ When I first joined Web3, I was simply curious about crypto and airdrops. I never imagined it would become such a meaningful part of my life. Thanks to Web3, I've made so many new friends. We started by talking about projects, daily tasks, and airdrops. But slowly, our conversations became about everyday life. We shared funny stories, complained about bad days, encouraged each other when things went wrong, and sometimes stayed up late just talking about random things. Some of these friends I've never even met in person, yet they've made my life feel a lot less lonely. And then came NeoSoul. My journey with @NeoSoulAI made me realize something else: Web3 wasn't just a place to make friends. It could also give me opportunities to earn extra income through my contributions. @Hansmahanakon @MollyApriltian That meant a lot to me. Being able to do something I enjoy, meet wonderful people, and earn something from my efforts made me appreciate this journey even more. Of course, Web3 hasn't always been easy. There have been disappointments, missed opportunities, and days when I felt exhausted. But when I look back, I feel grateful. Web3 gave me friendships when I needed connection, and NeoSoul showed me that my time and contributions could also have value. I may not have become rich from crypto, but my life has become richer in ways I never expected. And honestly, that's something I'll always treasure.

Mình mới dành thời gian đọc docs và xem flow của CONSO, thấy hướng đi khá khác so với mấy dự án mình từng thử. Phần Miner App mình quan tâm nhất. Reactor chạy session 24 giờ, kể cả khi đóng app vẫn tiếp tục tích lũy ZAP. Ngoài mining thụ động còn có mission invite bạn bè cũng tăng rate, và kết nối X mỗi cái thêm một ít boost. Điều mình thấy thú vị là mỗi CONSO BOT khi import vào app sẽ cho thêm 25% tốc độ mining so với base rate. BOTS còn đóng vai trò mobile node, giúp validate network và build reputation layer. Hiện mình đang theo dõi đợt free mint BOTS của @conso_xyz (5.555 con, mở 3 phase). Nếu được phân bổ thì sẽ import vào Miner App để tự trải nghiệm cách chạy Reactor + mission và xem reputation của mình được ghi nhận thế nào. Ai đang dùng Extension hoặc theo dõi CONSO thì nên xem qua phần docs trước khi mint, flow khá rõ.

CONSO BOTS. FREE MINT SẮP DIỄN RA! @conso_xyz 23/10 | 15:00 UTC (22:00 VN) 5.555 BOTS . Agentic Collectibles trên Sui 3 giai đoạn mint mở cùng lúc: GTD (48 giờ): Dành cho những người ủng hộ từ sớm. Impressions Raffle (2 giờ): Dành cho những người đã tích lũy impressions. Public Mint: FCFS đối với số lượng BOTS còn lại. Snapshot impressions: 12 giờ trước thời điểm mint. CONSO BOTS có gì đặc biệt? Hoạt động như Mobile Node trên Sui. Có khả năng mine ZAPs. Boost 25% cho mỗi BOT. Có thời gian lockup trước khi được giao dịch. Kiểm tra điều kiện & Mint: Kết nối tài khoản X + ví Sui -> Kiểm tra eligibility -> Claim theo phase của bạn. Free Mint, số lượng có hạn! Đừng bỏ lỡ cơ hội nếu bạn đủ điều kiện.

Spent some time going through the CONSO docs and the Miner App. @conso_xyz is building a Consumer Reputation Layer that turns activity across social, AI, gaming, and on chain into ZAPs and a portable Consumer Passport on Sui. Data is encrypted with Seal and stored on Walrus so you keep ownership. The Miner App lets you run a 24 hour reactor for passive ZAP mining plus missions. CONSO BOTS act as mobile nodes and give a +25% mining boost each . The Extension adds another layer by turning regular AI prompts into ZAPs while keeping everything local in the browser. Still early, but the idea of reputation that actually follows you instead of staying locked in each platform is interesting.

NEOSOUL BELIEVE PLAN. TEAM TOKEN UNLOCKS SHOULD BE BASED ON RESULTS, NOT JUST TIME @NeoSoulAI aims to build a sustainable ecosystem that can thrive for 100 years. @Hansmahanakon @YukiinWeb3yk I believe that to achieve this vision, team DNA token unlocks should be tied to real, measurable results that the community can independently verify. Here are my 3 proposed KPIs: 1. REAL PRODUCT REVENUE (40%) Revenue from Agents, Skills, and services on NeoTrade must grow by at least 15% compared to the previous evaluation period. Exclude internal transactions and refunded revenue. Verification through published revenue data and on chain transactions. 3. PAYING USER RETENTION (35%) At least 30% of first-time paying users must make another payment within 30 days. Exclude team wallets and invalid transactions. Verification through published payment records and user retention statistics. 3. AI AGENT PERFORMANCE (25%) Task completion rate must reach at least 90%. Serious error rate must remain below or equal to 2%. Verification through Agent performance reports and execution data. TOKEN UNLOCK RULES • The team must publish KPI results after each evaluation period. • All data and calculation methods must be transparent and independently verifiable by the community. • Team tokens should only unlock when all 3 KPIs meet their targets. • If the targets are not met, the tokens allocated for that period remain locked and are not automatically carried over to the next period. MY PERSPECTIVE A project aiming to survive for 100 years needs to prove three things: _Its products generate real revenue. _Its users keep coming back. _Its AI Agents perform reliably and effectively. When the team creates measurable value, it deserves to be rewarded. That is what DNA Proof of Commitment should mean under the Believe Plan. #NeoSoul #DNA #BelievePlan

A WORKING APPLICATION AND A WORKING CONTRIBUTION NETWORK ARE DIFFERENT REQUIREMENTS. Consider a browser based robotics collection platform. The simulator can load correctly. The controls can respond. A user can complete the task. That establishes an application level interaction. But a contribution workflow may also depend on services outside the simulation itself. @axisrobotics' official contributor journey includes completing a task and signing the contribution on Base. These are distinct operations. The simulator supports the interaction that produces the demonstration. The signing step supports the on chain contribution workflow. A successful local interaction should not be confused with confirmation that every downstream step has completed. This is a systems distinction, not a claim about a specific Axis failure. APPLICATION STATE What happened during the simulated task. CONTRIBUTION STATE What happened when the result entered the submission and signing workflow. For distributed Physical AI infrastructure, both states matter. A user facing workflow is only complete when its required stages have actually finished. #AxisRobotics

The Airdrop Checker is about to go live. @NeoSoulAI @YukiinWeb3yk @Hansmahanakon Please note: your airdrop will only be distributed if you successfully complete the checking process. No check means no airdrop. This step is mandatory, not optional. Once the checker is available, make sure you finish it as soon as possible and remind your friends to do the same so nobody gets left behind. Better to check early than miss out later.

Robotics deployment is becoming a network effect game. Not just better hardware. Not just smarter models. A game of who can connect the most real world experience. @PrismaXai A single robot in a lab learns slowly. Once robots start working in the real world different homes, warehouses, kitchens, factories every new deployment creates experience the others can use. Isolated robots stay limited. Connected ones get smarter together. The loop looks like this: More robots working in real places → more useful data → better models → robots that can handle more situations → more reasons to deploy the next ones. This is the same dynamic that made software platforms so powerful. Except now the feedback loop includes the physical world. The hard part has always been the data. Collecting diverse, high quality demonstrations across many environments is slow and expensive if every company does it alone. Most teams still treat data collection as something they have to build internally. That approach doesn’t compound. PrismaX is building the layer that turns this into a network. Instead of every robotics team collecting data in isolation, there’s a shared system where robots, human operators, and the data they produce reinforce each other. More participants → richer real world experience → better models for everyone using the network. The owner operator idea is what makes it scalable. People and organizations can run robots on real tasks and contribute the data that comes from it. They get paid for useful work. The network gets broader coverage of environments and tasks. Deployment stops being purely a capital and headcount problem. It becomes a participation problem. When this loop runs outside any single company, intelligence starts transferring across deployments instead of staying trapped in one lab or one customer. A model trained on one set of environments becomes useful in others. Human operators become force multipliers rather than permanent bottlenecks. The teams that win won’t necessarily have the most impressive single robot. They’ll be the ones that make every new deployment make the next one easier and more capable. PrismaX is building the service layer for physical AI so that network effect can actually form.

🧡 @NeoSoulAI Airdrop & $DNA TGE . What the Community Needs to Know @Hansmahanakon @YukiinWeb3yk After following the latest community updates, one thing is clear to me: The snapshot is complete, but the NeoSoul journey is entering a new chapter. Here's what stood out. 👇 1/ Airdrop distribution According to the meeting recap, distribution will prioritize registration time. Earlier participants are expected to receive tokens before later ones. This is only the first airdrop wave, with more distributions planned in the future. 2/ Important reminder for Genesis members and above If you hold any of these roles, make sure you've submitted the Google Form: Singularity, Soul-Architect, Soul-Sequencer I & II, Sync-Driver I & II, Immortal, Archon, Disciple, Genesis-OG or Genesis. ⏰ Deadline: October 14, 11:59 PM (UTC+8) 3/ Wallet accuracy matters Your submitted wallet must match the wallet associated with your contributions. The team plans to distribute tokens directly rather than requiring users to claim them. A separate allocation checking page is also expected. Sybil screening is part of the process, with duplicate or fraudulent wallets potentially disqualified. 4/ The $DNA token DNA is the announced token ticker. According to the meeting, tokenomics will be revealed on TGE day, and the team has signed NDAs with seven exchanges. Airdropped tokens are expected to have no vesting period. 5/ What happens after TGE? NeoSoul plans to introduce the Believe Program, with three tiers recognizing ecosystem contributors, real product users and community supporters. NeoTrade is also expected to transition from USDT top ups toward DNA based payments. And one important security reminder: There is no officially supported Neo or DNA staking program. Be careful with websites claiming otherwise. My takeaway: An airdrop recognizes the people who helped build the early community. But what matters next is whether that community continues using the products, contributing ideas and creating real value. From Discord roles to onchain agents, the next chapter is about participation beyond the snapshot. Looking forward to seeing how @NeoSoulAI develops after TGE. 🧡 $DNA #NeoSoul #TGE

Robotics Democracy: PrismaX Puts High Tech Ownership in the Hands of Ordinary People Robotics and Physical AI have long been controlled by big labs and well funded companies. Ordinary people were mostly just spectators. @PrismaXai is breaking that model. It is building a service layer where regular people can directly participate, contribute data, and earn money from skills they already have. You don’t need an engineering degree or an expensive robot. With just a camera or headset, you can record everyday tasks folding clothes, cleaning, cooking, organizing and get paid for quality egocentric videos. If you own a robot, you can become an operator, teleoperate it remotely on real tasks, and turn it into an income generating asset instead of something sitting idle. This is the Owner Operator model: individuals and small teams can register their robots, collect data, and join the network. Physical AI needs diverse, high quality real world data. PrismaX opens the door for thousands of ordinary people to help create it building a data bank with 120K+ episodes and 1,700+ scenarios. You’re not just earning. You’re helping shape the robots that will enter homes, factories, and cities. PrismaX isn’t promising robots will do everything. It’s building infrastructure so humans and robots work together and so ordinary people can own a piece of the value. That’s real robotics democracy.

Final episode of the NeoSoul newbie conference diary from Singapore is out! 🎬 Early morning hike with @dajotow, Ya Kun breakfast, then straight to the BNB Chain event, and wrapping the night with a proper lobster dinner 🦞 Best part? The PnL was already up over +1000 before dinner even started, so the whole meal was basically covered. Packed days, tired but totally worth it met so many cool people and picked up unexpected moments along the way. Two episodes still don’t fully capture how good TOKEN2049 felt. Huge thanks to everyone we crossed paths with and to @NeoSoulAI for making the Singapore trip special. Conference is done, but the Web3 journey continues. See you next time ❤️ @YukiinWeb3yk @Hansmahanakon

OPEN SOURCE INFRASTRUCTURE CHANGES WHO CAN EXAMINE A ROBOTICS WORKFLOW. A robotics platform can publish its results without exposing the tools used to produce them. Those are different levels of technical transparency. @axisrobotics has reported open sourcing components of its task generation and data cleaning pipelines. That creates an opportunity for developers to examine parts of the workflow rather than relying exclusively on descriptions of the final output. The value isn't that open-source software automatically produces better robot data. It doesn't. The value is that implementation details can become inspectable. Developers can study how a component is structured, identify assumptions, and evaluate whether its approach is relevant to their own work. A RESEARCH CLAIM EXPLAINS AN OUTCOME. AN OPEN SOURCE COMPONENT EXPOSES PART OF THE IMPLEMENTATION. Neither replaces the other. But together, they provide different ways to understand the engineering behind Physical AI. s.kaito.ai/BsTLXdu #AxisRobotics #Kaito

CONTACT DOESN'T ONLY MOVE AN OBJECT. IT CAN RESTRICT HOW THE OBJECT MOVES. Imagine a block resting freely on a table. It can slide across the surface. Now imagine a robot finger pressing the block against a wall. The available motion changes. The wall and the finger jointly constrain the object. The robot hasn't simply added another force. It has changed the set of physically possible movements. This is an important distinction in manipulation. An object may have several possible motions before contact. After contact, some of those motions become restricted. @axisrobotics collects simulated manipulation demonstrations through browser-based teleoperation. Those trajectories are records of actions occurring under physical constraints. That makes a broader question interesting: How does a policy learn which contacts create useful constraints for a task? This is a conceptual robotics question, not a claim about a specific Axis learning objective. The physical world isn't just a space where objects move. It's a system where interactions can change which movements are possible. #AxisRobotics #PhysicalAI

Molly’s clip is the kind of TOKEN2049 content that actually sticks. Not a booth pitch. Not a slide. She pulls someone new into the day, walks them through it, and explains NeoSoul in plain language: talk to the agent directly, how much of the capital each bet takes, what’s actually happening on site. Restaurant, car, event floor, then Hainanese chicken rice. That rhythm fits how @NeoSoulAI should be shown more than another stage talk. A short interview turns into something memorable because she knows how to carry a conversation. @YukiinWeb3yk @Hansmahanakon The best part of these conferences is never knowing who shows up next. She caught that, and NeoSoul landed right in the middle of it.

ONE PHYSICAL EXPERIENCE CAN EXIST AT MULTIPLE LEVELS Consider a robot putting an object into a container. At one level, there is detailed physical motion. Zoom out: GRASP → LIFT → TRANSPORT → PLACE Zoom out again: “Put the object into the container.” Same physical episode. Different levels of description. This matters when looking at the @axisrobotics stack. Its egocentric pipeline includes language and subtask annotations with frame level language alignment. Its Task Packages also distinguish scenarios from atomic skills such as grasp, place, push and rotate. That gives a useful way to think about robot experience: PHYSICAL MOTION ↓ SKILL ↓ SUBTASK ↓ TASK The source experience may be related. The useful representation depends on what you're trying to learn from it. Low level control needs detail. Skill learning needs reusable actions. Higher level reasoning may care about task structure. So “robot data” isn't necessarily one flat object. One physical experience can carry structure at multiple levels of abstraction. s.kaito.ai/BsTLXdu #AxisRobotics #Kaito