The Deng Exchange is open for ~24 hours, FCFS. 🪑 Trade in @DENGSonApe for credits; you must have some $APE on ApeChain for gas. 🪑 Use the credits to claim TMF Passes; you must have some $ETH on Robinhood for gas in your primary wallet. UGC Applications open Tuesday.

September wrapped for @axisrobotics. The number that matters: 5M trajectories on @base in September. Now past 6M. Still growing daily. The rest was proof, not announcements. Paper accepted to the IROS PWMS workshop. Presented in Pittsburgh on Oct 1. AXIS is building a community data engine for robot manipulation. Deployed policies collect the data that trains the next policy. Success improved from 22% → 52%. Open Axis Benchmark launched with @openroboto. Not a frozen test set. Each round pulls from: → 6,000+ tasks → 5.5M+ trajectories The score measures generalization, not memorization. The corpus just crossed 6M runs. Browser in. Policy out.

Some things are worth waiting for, especially when you feel like you could have a small place in the story for me and my wife, that little thing is a chair in @themutualfun🪑 Weve been following the journey, watching the building come together and waiting to see where our place might be i am not asking for anything special Just hoping that when the chairs are finally settled, there might be one waiting for us too @Kingpickle, my wife and I are still here patiently waiting So… do we get a chair in the family? 🪑

Everyone on the call wanted a TGE date. Simon didn't give one. And honestly that's the most bullish thing he could've said. The sequence is: orderbook → liquidity → traction → token. They're not skipping steps to satisfy a timeline. They're following the logic. @reya_xyz Season 2 is deliberately short. Fewer people competing for points right now means early users earn more. That's the stated incentive, not a guarantee, but the math makes sense. Golden Tickets coming in weeks, skip the queue, earn 10% of whoever you give it to. Forever. $135B settled in v1 before the token existed. The token comes after the product is proven. Not before. I've seen enough projects do it backwards to know which way I prefer.

Idle chatter x.lingyaoai.com/i/spaces/1nKOLQwWoAZGR

What if everyday machines could become productive assets on-chain? That’s exactly what @DualMintRWA is building > Laundromats > A/C units > Vending networks > And now arcade machines 1,590+ machines already running, with payouts going strong since May 2025 And now arcade revenue is live on @Solana No hype No imaginary yield Just real-world machines → real revenue → on-chain ownership Everything that generates revenue should have a token

Physical AI’s biggest problem isn’t just better models but the lack of quality data. That’s where @axisrobotics comes in. Axis is building a browser based robotics data platform where users can control simulated robots and create training trajectories. There are two main ways to contribute: > Pre-training: Create robot demonstrations >Post-training: Take over and correct the robot when it makes mistakes You need to Sign & Verify to make your contribution valid. Axis Points depend on: >Volume >Score > Difficulty > Diversity > Referrals The Hub has different tasks like Pick, Place, Stack, Rotate, Transfer, Arrange, Insert and Reach. Eligible Challenger Program users can earn 2–3× Axis Points on high-difficulty tasks. s.kaito.ai/VHqvYOr

agent projects stop at giving an agent an identity @termix_ai goes further with AACP An agent can find a service negotiate the terms fund the order onchain get the work back and settle payment all inside one protocol Discovery > Negotiation > Execution > Settlement Identity only gets an agent in the door Finished jobs are what build a record you can actually check 493952 jobs settled 464593 agents registered Last 24h 1311 new agents 10199 jobs settled I care more about jobs completed than agents registered Once agents can hire each other pay each other and earn reputation from real work it stops being a directory and starts looking like an economy

Everyone keeps asking the same questions about @housebets Wave 2, so here’s everything in one place - Is it a scam? Wave 1 paid out $2M+, and I got my payout too. - Do I need to connect my wallet? No. You only need to provide your wallet address. The wallet history is already public. - Why the $1 payment? It’s simply to verify that the wallet belongs to you. - Playthrough requirement? Just 1x. - Can I use multiple accounts? No. One claim per person. Multiple accounts can get disqualified. - Can I use an exchange wallet? No. Use your own personal wallet. If you participated in Wave 1, you’ll be re-checked based on your activity. My offer is worth $133 you can see it in the screenshot. housebets.com/r/NiftyHalo296

my friends are the most important people on Earth

The Franka Sim Dataset is the scale proof people skip. 207 manipulation tasks 50,000+ human demonstrations 60,000+ task and scene variants Seven scene categories All of it collected in the browser No lab rig in the room That is the browser plus retail user model @axisrobotics keeps pointing at. Not a demo day video. A dataset large enough to pretrain on, then correct, then pretrain again. September added a receipt. The Franka dataset family became the most downloaded open source simulated Franka manipulation dataset on HuggingFace. If crowdsourcing only produced junk, this set would not be the one labs keep pulling. The open question is transfer. Not whether people will play the claw machine.

Physical AIs biggest bottleneck isnt just better models its quality data thats where @axisrobotics comes in Axis is building a browser based robotics data platform where users control simulated robots and create training trajectories Current ways to contribute: - Pre training: create robot demonstrations - Post training: correct policy failures through takeover Valid contributions require Sign & Verify Points depend on -) - Volume - Score - Difficulty - Diversity - Referrals the Hub features tasks around skills like Pick, Place, Stack, Rotate, Transfer, Arrange, Insert and Reach Eligible Challenger Program users can earn 2–3× Axis Points on high difficulty tasks Points are not a token and airdrop allocation is not guaranteed the core thesis is simple Human feedback > better data > better robot policies > better Physical AI Join here -) s.kaito.ai/gsTfJU8

Gm Gn Everyone AI agents are moving beyond chat. They’re starting to deliver actual services. A new video production agent has joined @termix_ai on agent.family. One prompt. One finished promo film. Just provide: → Your brand → Your message → Your vibe → Logo and copy The agent handles the workflow: Prompt → Visuals → Music → Final cut. Here’s what makes this interesting. TermiX isn’t just building a directory of AI agents. Through its AACP protocol, it aims to make agents discoverable, hireable, and capable of completing jobs with verification, reputation, and on-chain settlement. Imagine hiring an AI agent for a promo video today, then finding agents for research, coding, design, or data collection through the same infrastructure. That’s the bigger idea: an economy where AI agents can offer services, get hired, deliver work, and get paid. Check out the finished video: agent.family/listing?id=cmu Explore the ecosystem: agent.family

The interesting part of prediction markets isnt the prediction its what happens when conviction gets a price thats what caught my attention about @xomarket XO is building Conviction Markets where people can turn their view on a future event into a tradable position you choose Yes or No, enter at the price you believe is right, and the market continuously reflects where collective conviction stands Shares trade between $0 and $1 If your outcome wins, each winning share settles at $1 and you don't necessarily need to wait for resolution positions can be sold earlier at the prevailing market price depending on liquidity and fees but the interesting part goes beyond trading eligible users can create markets by defining the question, outcomes, resolution rules data source and resolution date XO also has Conviction Points (CP), rewarding users for holding positions, with stronger incentives for views that differ further from the current consensus the model is simple -) View - Position - Price Discovery - Consensus - Resolution XO isnt just trying to answer what happens next its creating a way to measure how strongly people believe it will happen thats a much more interesting market to watch Join here -) s.kaito.ai/lcEEQRm

prediction markets are built around what's already popular elections price targets big games but the questions I actually want to trade are the weird ones will this small project ship before the deadline will this creator hit that milestone this month stuff no big platform would ever list @xomarket lets you create that market yourself you write the question others take positions and the price moves as new info comes in adaptive liquidity keeps trading smooth AI + decentralized arbitration handles resolution so the people closest to a niche topic are the ones who get to shape the market around it what's one market you wish existed right now? s.kaito.ai/ni7oPXx

GM GN CT TermiX is starting to look less like an AI agent directory and more like an actual marketplace. The latest numbers caught my attention: $19.68M+ settled through on-chain escrow 431K+ agents indexed 370K+ jobs completed $393K+ protocol revenue And the interesting part is the scale of the transactions. The average job is only around $53. That means this isn’t just a few large contracts inflating the numbers. It’s thousands of smaller agent-to-agent transactions happening onchain. The flow is simple: discover → quote → escrow → deliver → challenge → settle. And now a new video production agent has joined agent.family. One prompt → brand + message + vibe + logo/copy → finished promo film. That’s a good example of where this is heading: agents aren’t just being listed, they’re becoming actual services people can hire. With BNB Chain’s new Set and Earn campaign also putting TermiX in front of more users, the focus is shifting from “what could the agent economy become?” to actually testing it. Video agent: agent.family/listing?id=cmu That’s the part I’m watching with @termix_ai.

Most people look at the USDC number on an @xomarket reward pool and stop there. The number that decides what you earn is how far from the midpoint you quote. The docs score each order on size, proximity to the midpoint, and time resting. Proximity weight is quadratic: ((D − d) / D)². On a pool with a ±10¢ max: 1¢ away → 0.81 2¢ → 0.64 3¢ → 0.49 5¢ → 0.25 7¢ → 0.09 10¢ → 0 Read it as a capital-efficiency table. To match the score of a 1¢ quote, a 3¢ quote needs about 1.65x the size. A 5¢ quote needs 3.2x. A 7¢ quote needs 9x. Quoting wide to avoid fills isn't free, because you pay for it in capital that has to sit on the book. The docs' own example: Maya quotes 100 shares on both sides at 49¢/51¢ and rests 12 hours. Lee quotes 100 shares on one side at 49¢ and rests 24 hours. Lee's order sat twice as long and he still lost, 648 to 972. Maya took 60% of that day's pool, because that pool pays Single 1× / Both 3×. Quoting both sides is worth triple. Fine print that costs people their score: Cancel and repost, and the resting clock restarts (30 seconds by default) After a partial fill the clock keeps running on the remaining shares, but if what's left drops below min size, it stops scoring The midpoint comes from the best qualifying bid and ask. Orders under min size don't set it, and with no valid two-sided book nobody scores that interval Pools near 0¢ or 100¢ can require both sides, since quoting the obvious side there is easy to game Epochs run 00:00 to 00:00 UTC, and parameter changes only apply from the next epoch An order doesn't have to fill to earn. But it can fill, and then you hold a real position. The docs say plainly that rewards may not cover that loss, so price your distance against how much of that position you'd want. Read the pool card (Max, Min size, USDC pool, CP pool, eligibility) the way you'd read a fee schedule. It tells you the strategy.

Today my Sorsa score went up by +1 Total score: 364 Whats your sorsa score?

a project with unlimited scammer prove me wrong

Good day everyone @agenticscredit scanned every trade ever placed on Polymarket and found 500,000 wallets already clearing the qualifying bar. No forms, no stake, no PnL screenshots. Just history long enough and clean enough to underwrite. The number is the headline, but the quieter part is what it implies. Most of those wallets never set out to build a credit score. They just traded, and the record was sitting there the whole time waiting to be read. That's the shift. Reputation stops being something you pitch and becomes something your history already proves. Clearing the bar isn't funding yet, the credit path still runs through a waitlist, but having a readable record is where it starts. t.me/Agenticscreditbot?start…