Pinned Tweet
What actually changed when Rayls Sovereign replaced the Privacy Node?
I read the current docs. It's a re-architecture, not a rename.
GETH Became RETH !
Clique proof of authority became Axyl consensus. The Key Management Module became the Cryptographic Trust Suite
The Relayer moved from MongoDB to PostgreSQL.
The topology is hub and spoke. Each institution runs its own spoke, a ledger plus the services beside it. Every spoke connects to one shared Private Network Hub, and no spoke talks to another directly. The Hub routes encrypted messages and can't read their contents.
Thread on what each change means, and what the docs don't reconcile.
Blaze Torc retweeted
She can ask for AI chips, offer all kinds of deals, but my Tangem secret codes are staying with me. 😂
Nice try, but nobody gets access to my keys🔐
My crypto... My keys.... My control.
Join now : t.me/TangemIndia
@Tangem #Tangemindia
Blaze Torc retweeted
Is Mark Zuckerberg the last true wife guy standing? What makes him stand out compared to other famous men?
Is Mark Zuckerberg the last true wife guy standing? What makes him stand out compared to other famous men?
Blaze Torc retweeted
Atleast once in your lifetime, you'll find someone you have insane chemistry with and it is very important to know you cannot be with them.
Atleast once in your lifetime, you'll find someone you have insane chemistry with and it is very important to know you cannot be with them.
Blaze Torc retweeted
What actually changed when Rayls Sovereign replaced the Privacy Node?
I read the current docs. It's a re-architecture, not a rename.
GETH Became RETH !
Clique proof of authority became Axyl consensus. The Key Management Module became the Cryptographic Trust Suite
The Relayer moved from MongoDB to PostgreSQL.
The topology is hub and spoke. Each institution runs its own spoke, a ledger plus the services beside it. Every spoke connects to one shared Private Network Hub, and no spoke talks to another directly. The Hub routes encrypted messages and can't read their contents.
Thread on what each change means, and what the docs don't reconcile.
Replying to @blazetorc @RaylsLabs
5/6
How messages stay private between institutions:
ML KEM establishes a shared secret between two institutions. HKDF SHA3 derives a symmetric key from it. AES 256 GCM encrypts the payload.
Rayls describes the key encapsulation as post quantum on its Public Chain product page. The Hub stores and forwards ciphertext it can’t read.
What leaves the institution is an encrypted payload plus a Merkle proof. The Hub can verify the proof without reading the payload. Plaintext balances, counterparties and contract state stay on the institution's own hardware or cloud tenancy.
WHAT DIDN'T CHANGE: The Hub still runs Besu with proof of authority and is unaffected by the Axyl rollout. So Axyl's throughput figures describe the Sovereign ledger, not the cross institution path.
1
33
6/6
What I could verify, and what I couldn't.
The Sovereign intro page cites 15,000+ TPS and sub second finality. The benchmarks page gives two different pictures:
Prototype, 4 node committee:
250,000 tx/s, 166 ms consensus latency, 241 ms end to end
Rayls Testnet: 26,243 tx/s consensus, 0.5 s block time, finality above 1 second.
The prototype used 512 B transactions, 500 KB batches and a 50 ms maximum batch delay.
My arithmetic: 250,000 x 512 B is about 1 Gbit/s of raw payload, and at 500 KB per batch that's roughly 250 batches a second, so batches fill by size, not by the 50 ms timer.
The docs don't say which figure applies to a production deployment, or how finality is defined. These are different environments, so it isn't necessarily a contradiction, but it isn't reconciled.
On production: Rayls reports 30+ institutions and names Nùclea (live since June 2024, on the earlier Privacy Node platform) and XP (the USDXP stablecoin).
Sources: docs.rayls.com (Sovereign, Architecture, Key management, Axyl benchmarks), and Rayls "Introducing Rayls Sovereign" blog post.
1
31
Want A $1000 Poster Output?
This is absolute fire 🔥
AI slop design is everywhere. Good design is still rare.
The problem isn't the AI tool. It's expecting creativity from a single prompt.
📝Here's how I combine human taste, design references, and AI to create better visuals:
1. I don't start by asking AI to create a design from scratch.
2. First, I visit platforms like Pinterest or Behance to explore designs created by others.
3. I search for what I need using simple keywords like "X banner", "X cover photo", or "LinkedIn poster".
4. I go through different designs and find one that closely matches what I have in mind.
5. I don't plan to copy the entire design. I use it as a reference and starting point.
6. Then I think about what I can improve, remove, replace, or add to make it better.
7. Sometimes, I explore more designs to find other elements, backgrounds, typography , colors, or styles that could work well together.
8. I think about how these different elements can be combined naturally into one design.
9. Once I have a clear idea, I upload the reference image to an AI image generation tool.
10. I explain exactly what I want, step by step, including what should stay the same and what needs to change.
11. If my requirements are complicated, I first ask AI to organize my ideas into a clear, detailed image generation prompt.
12. I read the generated prompt carefully instead of using it immediately.
13. I check whether the prompt correctly explains the reference image, my intended changes, and how everything should fit together.
14. If something is missing or unclear, I ask AI to improve the prompt.
15. Once I'm satisfied with the instructions, I use the final prompt and reference images to generate the design.
16. I review the output, identify anything that looks wrong, and make further adjustments if needed.
17. I repeat this process until the design looks close to what I originally imagined.
🔑 The idea is simple: Don't let AI do all the thinking.
🔑 Use your own creativity to explore, select, combine, and improve ideas.
🔑 Let AI help you turn those ideas into a finished design.
💌 And finally, huge respect to the original designers and creators whose work inspires us.
💌 Their creativity, attention to detail, and countless hours of hard work are what make these designs possible in the first place.
💌 AI may help us create faster, but the creative minds behind the original work deserve the real credit.
5
71
Blaze Torc retweeted
Hey @4Dlabs_Official i just redesigned the new Season 1 announcement poster.
Hope the team will see this and post there thoughts here !
How is it guys? Rate it in comments below ;)
@4Dlabs_Official #edit #design
⚡ 4D Labs Season 1 is coming.
A new way to participate in the 4D Labs ecosystem is about to begin.
🧩 Complete new missions
🌐 Contribute to the community
🦾 Help build the real-world data layer for Physical AI
Season 1 goes live soon.
Get ready.
#4DLabs #Season1 #PhysicalAI #EmbodiedAI #RealWorldData










