Most new contributors on @axisrobotics tend to stick with pre-training tasks because they feel more straightforward. But the real opportunity might be in post-training. Pre-training teaches a robot how a task should be done. Post-training shows where the trained policy actually fails, then captures the correction at that exact moment. That makes the data much more targeted. It also explains why 660 corrections were narrowed down to just 161 verified snippets. Useful correction data is hard to collect, but when it’s valid, it points directly at a model’s weakness. So if post-training feels harder, that’s kinda the point. The harder tasks are often where the most valuable gaps are.

Hype is easy. Research is harder to fake. That’s one reason I’m paying attention to @axisrobotics. The team combines robotics researchers from places like Berkeley, CMU, UCLA and NTU with people who have experience scaling products. Then there’s the academic collaboration behind Axis Dataset V1 and the publicly available research. The interesting part isn’t the university names. It’s that the work can actually be examined and challenged.

Just put in my application for @FungoLabs. What got me interested is the way they’re approaching NFTs with privacy built in from day one, rather than trying to add it later. The @zama FHE angle makes it even more interesting since encrypted data can still be used by smart contracts. Still early, but definitely one I’ll be watching closely. 👀

Hey man, $POLY speculation is getting loud again. Shayne’s latest Token2049 tease has people connecting the dots, and POLY is coming is already making the rounds. But there’s still no confirmed TGE date or airdrop details. I’m watching it too, but I’m not assuming an announcement is happening this week. For now, I’m just staying active on @Polymarket and waiting for something official. The interesting part isn’t whether the timeline can hype $POLY. It’s when Polymarket actually confirms the next step.

A lot of Physical AI projects talk about building the “data layer.” What caught my attention with @axisrobotics is that there’s a roadmap behind the claim. Some milestones are already completed, while others are still targets. Q1: core data loop validated Q2: Franka Sim Dataset shipped with 207 tasks and 50K+ demonstrations Q3: V2 closed loop + global rollout Next: large-scale DAgger data and $1M+ ARR target That last part is especially important. If Axis can turn its growing data network into real demand from robotics companies, the thesis becomes much stronger. For me, the interesting part now isn’t just whether the dataset keeps growing. It’s whether the data engine can become a real business.

One thing I find interesting about @axisrobotics is how they approach data verification. Instead of asking buyers to simply trust the dataset, each signed trajectory can have a unique Data ID recorded on Base, linked to the task and contributor. That creates a clear provenance trail. You can see where the data came from, who produced it, and what task it belongs to. For Physical AI, that kind of transparency matters. The data itself is valuable, but being able to verify its origin makes it even more useful.

The scale behind @axisrobotics is worth looking at. 94K+ contributors, 2.1M+ trajectories, and 1,600+ published tasks show how much data the network has been building. But raw numbers only tell part of the story. What really matters is how that data is collected, verified, and used to improve robotic learning. For Physical AI, it’s not just about gathering more data. It’s about building a dataset with enough variety and coverage to help models handle real-world tasks. That’s the part of Axis I’m watching closely.

A typical day trading on @Polymarket : Checking ETH/BTC/BNB markets, tracking price movements, waiting for the right setup, then making a decision. The hardest part isn’t clicking buy or sell. It’s staying patient and not forcing a trade when the setup isn’t there. Discipline matters more than speed.

Epoch 2 just started for @axisrobotics × @KaitoAI. If you missed the first round, this is a fresh reset. 30 days to create, contribute, and build your presence around Physical AI. 0.25% of $AXIS supply is still allocated, with your top 6 posts counting from up to 10. Fresh epoch. Fresh grind. Let’s see what this round brings.

Most people assume robotics data needs to be nearly perfect before it becomes useful. But @axisrobotics is exploring a different approach. Instead of relying only on a small group of expert operators, the idea is to collect diverse demonstrations from a much larger contributor base. Different people approach the same task differently. Some movements are efficient, others are imperfect, but together they can provide a broader range of experiences for training. Of course, more data alone doesn't guarantee better results. Data quality, diversity, and successful learning all matter. What makes Axis interesting is the focus on scaling robotic data collection beyond traditional, controlled environments. Physical AI won't advance through better models alone. The data behind those models matters just as much.

Hello mate. If you’re farming @axisrobotics points, more submissions doesn’t automatically mean more points. The system looks at multiple things: • Volume • Submission quality • Task difficulty • Diversity across scenarios • Referrals based on actual contribution One important detail: repeating the same task up to 5 times uses the average score, so a few weak attempts can actually hurt your points. Harder tasks also carry more weight than repeatedly doing easy ones. So the better strategy isn’t simply to do more. Do harder tasks. Keep your quality high. Try different scenarios. Avoid unnecessary repetition. On Axis, meaningful data beats empty volume.

WL application is now live for @FomiesNFT. Been keeping an eye on this one, and the waitlist is finally open. If you’re interested, now is the time to secure your spot. Don’t wait until the last moment. 👀 Join: fomies.family/?ref=32fce6990…

Gud luck. fomies.family

$AXIS community sale just crossed $1M+ in commitments. Already oversubscribed. @axisrobotics is now generating 100K+ Physical AI data trajectories every day. Commitment window stays open until Sep 28, 13:00 SGT. Pro-rata allocation. Unaccepted USDC gets refunded.

My SmartX trader type: “The Alpha Caller.” “Three hours of research. Two-word thesis: send it.” Find yours in 6 questions. Join the waitlist for up to 50% off fees and a share of 10,000 USDT in rewards. smartx.io/waitlist/?invite=p…

Still waiting on that approval from @BITFOOTS_ Applied 2 days ago and I’ve been staying active, supporting the project, and creating along the way. But the Approved status is still missing. @BITFOOTS_ team, if you see this, please take a look at my application and artwork. Small creator, but real support. Anyone else still waiting?

The countdown is on for @Zcash. Tuesday, Sept 22 at 5 PM UTC. The plan: → 6,000 hackers on $ZEC → Free mint → Minting through @zec_bit launchpad → Keep around $7 in ZEC for network fees → Wallet checker coming soon A new drop for the cypherpunk crowd. Cypherpunks write code.

overthinkers, we saved you a seat👀 whitelist raffle registration is now open. follow the ant. 🐜 fomoater.com