ummaima retweeted
Spain's data regulator published its first breach report tied to an AI agent on September 15. It named no company, product, or cause. It named where to look: prompts, logs, conversation histories, agent memory. Memory is now a reportable surface.  cypro.co.uk/insights/cyber-b…
27
36
121
4,725
ummaima retweeted
Gud luck. fomies.family
15,766
13,596
15,567
371,369
AI is moving from systems that simply answer to systems that can participate. That’s the idea behind @ritualnet’s approach to autonomous intelligence A long-lived AI agent needs more than a model. It needs identity, execution, pub àqa1memory privacy, coordination, and the ability to operate over time. Ritual is building these capabilities directly into its blockchain infrastructure The goal is to let agents interact with AI models, access information, schedule execution, and operate onchain. One interesting piece is native AI execution. Agents can call AI models through onchain primitives, with computation handled through privacy-preserving infrastructure. Then there’s autonomous scheduling Instead of waiting for a human or external keeper to trigger every action, agents can schedule recurring or conditional execution Ritual also focuses on confidential computation. Keys, prompts, and sensitive strategy can remain protected while the system still provides cryptographic guarantees around execution. Put these pieces together: AI + execution + privacy + scheduling + capital And AI agents start looking less like simple applications and more like persistent onchain participants That’s what makes Ritual interesting to watch The goal isn’t just to put AI on a blockchain It’s to build infrastructure where autonomous intelligence can actually operate
4
19
110
Pw
AI going onchain is more than connecting models to blockchains The bigger question is What infrastructure does AI need to operate autonomously? That’s where @ritualnet gets interesting Ritual is building infrastructure for autonomous intelligence enabling AI agents to reason coordinate transact and execute onchain The goal is to make AI a native participant in the onchain economy One key piece is native AI computation Ritual is designed to give developers access to expressive compute and AI capabilities directly within onchain applications Autonomous agents also need more than intelligence They need the ability to hold keys manage permissions, schedule actions, and operate with programmable budgets That’s an important primitive for agent-driven applications Privacy is another major part of the architecture Ritual focuses on combining private computation with cryptographic verification: Private thought Public proof The computation can remain private while its result can be verified. This opens up new possibilities → Autonomous DeFi → AI-native applications → Agent-to-agent transactions → Prediction markets → Machine-native economies All built around programmable intelligence The interesting shift is not simply How can blockchain use AI It’s What infrastructure does autonomous intelligence need to exist onchain That’s the problem Ritual is building aroaroun The future of onchain applications may not be limited to users clicking buttons It could include autonomous software that can reason coordinate, transact and execute continuously Ritual is building toward that future.
12
210
Ethereum can have powerful validators and strong consensus, but if data moves slowly, the whole network feels it That’s where @get_optimum is taking a different approach Traditional GossipSub relies on peers forwarding messages across the network It works, but the same data can be transmitted repeatedly, creating extra bandwidth usage and propagation overhead. Optimum’s mump2p approaches this differently with RLNC Random Linear Network Coding. Instead of treating a block as one complete message, data can be split into coded pieces that move through different paths The interesting part is that intermediate nodes can recoded the data they receive and keep forwarding useful information That means the network doesn’t always need to wait for the exact same pieces to arrive This can make data propagation more flexible Multiple paths Coded shards Smarter forwarding Less redundant transmission All while working toward the same goal more efficient blockchain networking Ethereum’s performance isn’t only about execution or consensus How quickly blocks and attestations travel between validators matters too Optimum is exploring the networking layer where those milliseconds can matter The bigger idea Don’t just build faster blockchain infrastructure. Build better roads for the data moving across it That’s the problem @get_optimum is working on with mump2p
Optimum is taking its infrastructure conversation to KBW and Token2049. But beyond the events, there’s a bigger idea behind the work: improving how blockchain data moves across networks. One example is mumP2P Traditional P2P networking mainly focuses on connecting nodes and exchanging data. mumP2P takes a different approach — focusing on how blockchain data moves through the network efficiently as networks become larger and more demanding. Instead of simply adding more peer-to-peer connections, mumP2P focuses on the movement of data itself. The goal is to help information propagate through the network with less unnecessary overhead. Blockchain nodes constantly deal with: → Bandwidth → Latency → Memory → Changing network conditions mumP2P is designed around making better use of these resources. And this is where infrastructure becomes important Faster data propagation can help validators receive information sooner, reducing delays between when data is produced and when it reaches the rest of the network. Now, the team is bringing these infrastructure conversations to the real world. Sept 28 — KBW @kentlinyy joins @ethereumkoreaio’s Staking Infra panel. Sept 29 — CODED Optimum hosts its signature CODED event, featuring a keynote from @MurielMedard and speakers from 17 top projects More event details and sign-up info are coming Oct 6 — Token2049 Optimum is cohosting an event on agentic trading with @ambr_io. Then on Oct 7, the team will be attending DAS Asia From optimizing blockchain data movement with mumP2P to discussing the next generation of infrastructure at KBW and Token2049, Optimum is continuing to build around one important layer: the infrastructure that makes decentralized networks perform
2
16
138
ummaima retweeted
What if the crowd is wrong? And what if you’re the one person who can see it? That’s where @acepyr gets interesting. Meta Markets turn collective belief into something you can actually measure then challenge. Would you bet against the crowd?
Everyone knows what they think. Almost nobody knows what everyone else actually thinks. That's the interesting part. @acepyr isn't just asking: “What will happen?” It's also asking: “What does the crowd actually believe?”
Paid partnership (ad)
110
22
289
1,497
ummaima retweeted
IROS is days away. Here is the question @castorhat will answer on the Tech Talk Stage: What makes good robotics data? Tuesday, September 29, 3:15 PM. @ieeeiros #IROS2026
18
8
97
2,766
ummaima retweeted
On September 4, an oracle on Starknet briefly reported USDT at about $2.04, and Vesu liquidated 47 positions in the next 109 seconds. Pragma published its post-mortem on September 14. The data flowed exactly as designed and the delivered number was still wrong.
48
58
181
8,001
ummaima retweeted
10 X GTD WL @Kageuri_com Follow @Kageuri_com x @ModeXeth Like & Retweet send wallet👇
1,096
922
1,140
19,454
Optimum is taking its infrastructure conversation to KBW and Token2049. But beyond the events, there’s a bigger idea behind the work: improving how blockchain data moves across networks. One example is mumP2P Traditional P2P networking mainly focuses on connecting nodes and exchanging data. mumP2P takes a different approach — focusing on how blockchain data moves through the network efficiently as networks become larger and more demanding. Instead of simply adding more peer-to-peer connections, mumP2P focuses on the movement of data itself. The goal is to help information propagate through the network with less unnecessary overhead. Blockchain nodes constantly deal with: → Bandwidth → Latency → Memory → Changing network conditions mumP2P is designed around making better use of these resources. And this is where infrastructure becomes important Faster data propagation can help validators receive information sooner, reducing delays between when data is produced and when it reaches the rest of the network. Now, the team is bringing these infrastructure conversations to the real world. Sept 28 — KBW @kentlinyy joins @ethereumkoreaio’s Staking Infra panel. Sept 29 — CODED Optimum hosts its signature CODED event, featuring a keynote from @MurielMedard and speakers from 17 top projects More event details and sign-up info are coming Oct 6 — Token2049 Optimum is cohosting an event on agentic trading with @ambr_io. Then on Oct 7, the team will be attending DAS Asia From optimizing blockchain data movement with mumP2P to discussing the next generation of infrastructure at KBW and Token2049, Optimum is continuing to build around one important layer: the infrastructure that makes decentralized networks perform
9
24
312
Pw
Korea Blockchain Week Something interesting is coming to Korea Blockchain Week @kentlinyy from @get_optimum will be joining Ethereum Korea’s Staking Infra panel on September 28 One topic worth watching mump2p The Network Layer Matters Ethereum performance isn’t only about execution How quickly blocks and attestations move between nodes matters too That’s where Optimum’s mump2p networking layer comes in RLNC for Data Propagation Optimum uses RLNC (Random Linear Network Coding) to make blockchain data propagation more efficient and resilient. The goal is simple: better propagation with more efficient network resource usage DSRV × Optimum DSRV has joined as a new mump2p testing partner Their experience with institutional staking infrastructure adds another interesting dimension to Optimum’s networking research CODED@KBW On September 29 Optimum is also co-hosting CODED@KBW with Google Cloud Four Pillars LINE NEXT and Auros The event will explore scalable blockchain infrastructure stablecoins payments and AI agents Better Data Movement Different conversations same underlying theme Blockchain infrastructure needs efficient data movement mump2p + RLNC are tackling an important layer of that problem @get_optimum
1
10
269
20 have already been raffled and this is the last giveaway from the Hazels project👇 GIVEAWAY – hazels🔥 (10 WL ) @saka_sol × @0xhazels Winners picked within 10 hours 1. Follow @0xhazels / @saka_sol 2. Like and retweet 3. Drop your EVM wallet address
213
185
239
4,519
Physical AI is moving beyond the lab For robots to work reliably in the real world they need more than powerful hardware they need real-world data human interaction, and better training infrastructure. That’s the space PrismaX is building in. Its Robotics Data platform connects real-world robot demonstrations with AI development, helping teams collect and use data across different tasks, environments, and robotic systems. The scale is already notable: → 5,000 hours/month capacity → 1,700+ deployed scenarios → 120K+ episodes delivered More real-world demonstrations means more opportunities to train physical AI systems. PrismaX also brings humans into the loop. Through its VLA Foundry, operators can capture demonstrations from real robotic tasks and contribute valuable training data to the ecosystem. Humans demonstrate. Robots learn. Models improve Data quality is just as important as data quantity. PrismaX’s Verify Quality system lets contributors evaluate robot training data and build a reputation around the quality of their work. That creates an important feedback layer for physical AI The ecosystem also reaches hardware. PrismaX’s Robot Fleet brings robotic platforms into a marketplace, including systems from AGILEX, Airbot, and I2RT. Hardware + operators + data + AI can work together instead of staying in separate layers. The bigger vision is a connected ecosystem for physical AI Real-world data better training smarter robots more real-world applications. PrismaX is working on the infrastructure that can help connect those pieces Physical AI needs intelligence that can operate in the physical world
🔽 PRISM From Hardware to Data @PrismaXai is building more than a robotics platform. It connects robots → real-world interaction → data → verification intelligence into one connected pipeline Hardware Operation Data Verification Intelligence 🔽 01 — HARDWARE For Physical AI hardware is only the starting point PrismaX’s Robot Fleet connects validated robots with the data-collection workflow while supported robots can be registered on the platform. 🔽 02 — DATA COLLECTION Robots perform real-world tasks, generating valuable robotics data through their interactions with physical environments. PrismaX enables operators to upload robotics data for VLA training, including MCAP files and videos mapped to task scenarios. 🔽 03 — VERIFICATION Collecting more data isn’t enough. PrismaX’s Verify Quality layer focuses on reviewing and scoring data quality, with validators evaluating training data The goal is not just more data but better data 🔽 04 — DATA STANDARDS Robotics data quality depends on factors like task selection sensors, embodiment and the collection process PrismaX focuses on establishing repeatable standards for producing and validating training-grade robotics data 🔽 05 — INTELLIGENCE Verified data has a bigger purpose than simply creating datasets It can support the training and improvement of robotics models The loop becomes Robots Data Verification Models Better Robots 🔼 06 — THE BIGGER PICTURE Physical AI needs more than capable models. It also needs high-quality real-world data PrismaX is working to coordinate robots, human operators data and verification within one ecosystem 🔽 07 - The bigger idea is simple Hardware creates interaction Interaction creates data. Verification creates quality Quality helps build better intelligence. Hardware Data Verification Intelligence Physical AI That’s the pipeline PrismaX is building @PrismaXai
4
11
112
ummaima retweeted
Robotics Data is open in limited release. Usable, diverse, proven demonstrations for the teams building physical AI. Teleoperation and egocentric episodes, reviewed to one standard, growing every month. Explore the data and request access: prismax.ai/#plans
52
25
174
15,098
ummaima retweeted
TradFi latency optimization is straightforward, colocate or run fiber direct to the exchange. In DeFi you can't do that because you don't know the path that data will need to travel. Our CEO @MurielMedard recently went on A Global Tech Podcast to explain how Optimum solves this.
42
44
206
15,031
ummaima retweeted
Is the bitcoin:native bull market starting?
75
56
209
306,990
The Signal in the Static Information is everywhere. Opinions are everywhere. But knowing what people actually believe is much harder That’s where @acepyr takes an interesting approach Acepyr is built around prediction markets where the market price becomes a forecast When people put their conviction behind a position that price becomes a signal of what the crowd expects But Acepyr goes further with Mirror Instead of asking What will happen Mirror asks What does everyone actually believe The idea is simple but powerful People vote privately, while traders take positions on where that hidden crowd belief will land The market is essentially trying to read the room. That creates a different kind of signal Not just the loudest opinion on social media Not just the trend getting the most attention But a measurable view of collective conviction And that’s what makes @acepyr interesting to watch It isn’t only about predicting outcomes It’s about putting a price on what people believe and creating a record that can be tested against reality In a world full of noise the interesting question becomes Can you find the signal That’s the idea behind Acepyr
Acepyr is building a different kind of market Instead of only asking what will happen, it explores what people actually believe turning opinions and conviction into measurable market signals With live 5-minute markets, leaderboards, referrals and upcoming meta-markets the ecosystem keeps expanding Still early Still pre-mainnet $ACEPYR @Acepyr
5
1
17
168
ummaima retweeted
Replying to @X
Timelines are about to drastically change from here on
30
30
87
20,950
Korea Blockchain Week Something interesting is coming to Korea Blockchain Week @kentlinyy from @get_optimum will be joining Ethereum Korea’s Staking Infra panel on September 28 One topic worth watching mump2p The Network Layer Matters Ethereum performance isn’t only about execution How quickly blocks and attestations move between nodes matters too That’s where Optimum’s mump2p networking layer comes in RLNC for Data Propagation Optimum uses RLNC (Random Linear Network Coding) to make blockchain data propagation more efficient and resilient. The goal is simple: better propagation with more efficient network resource usage DSRV × Optimum DSRV has joined as a new mump2p testing partner Their experience with institutional staking infrastructure adds another interesting dimension to Optimum’s networking research CODED@KBW On September 29 Optimum is also co-hosting CODED@KBW with Google Cloud Four Pillars LINE NEXT and Auros The event will explore scalable blockchain infrastructure stablecoins payments and AI agents Better Data Movement Different conversations same underlying theme Blockchain infrastructure needs efficient data movement mump2p + RLNC are tackling an important layer of that problem @get_optimum
5
12
187
Pw
1/4 gmum One thing that stands out while learning about @get_optimum is how quiet the real constraint often is Most conversations around blockchain performance focus on execution, scaling, or new virtual machines But underneath those layers is a simpler question How efficiently can data move from point to point 2/4 When data delivery is inefficient everything above it inherits the cost Latency increases, bandwidth gets wasted and block propagation becomes harder to optimize That makes the network layer an important part of blockchain performance 3/4 This is where Optimum’s approach becomes interesting Using techniques like RLNC Optimum is working toward reducing redundant traffic and making data propagation more efficient. The goal isn’t to change consensus It’s to improve how information moves between nodes 4/4 The deeper I look, the clearer it becomes Blockchain performance isn’t only about processing data faster It’s also about moving data smarter from point to point with less waste That’s a layer worth paying attention to @get_optimum
1
9
234