@eightlends@TheBlockCo Yea I think clearing systems do exist at scale in financial systems but building one with blockchain rails with disparate blockchains, finalities, token standard is a unique challenge. Prize is same everywhere : incredibly scalable efficiency.
We have been working on this for a while and very excited to have such an incredible investor bench that allows us to bring a product we are proud of and is genuinely a technological leap compared to what’s available in crypto today. More soon !
We are incredibly proud to announce ZeroDelta publicly and a huge thank you to our investors and supporters over the past 2.5 years. 🫡
We will be disclosing more about our vision and about the product soon!
sad not sure what is happening but fable is really behaving stupidly and wasting a lot of time meanwhile gpt 5.6 oneshots best answer, better creativity and finds holes in fable 5's strategies.
@theo Instead I created a skill where I instruct current Fable on Claude code to replace claude.md for specific tasks and invoke codex on the fly to get some work done. Also built a similar skill to invoke another account on Max keeping my entire environment intact.
@theo Instead I created a skill where I instruct current Fable on Claude code to replace claude.md for specific tasks and invoke codex on the fly to get some work done. Also built a similar skill to invoke another account on Max keeping my entire environment intact.
@quxiaoyin Because trust is built over decades and based on precedents. When you have no IP laws enforcement paths and history of copycat products saving 100X could mean life or death of your company. Don’t forget how these models are cheap in the first place : distillation attacks.
One easy way to gain 5X improvements on your Claude setup is to let Fable5 evaluate your subagents and rewrite them after understanding the intention of each of those subagent.
As we are waiting for fable 5 to come back I figure it was better to spend time on recharging and resting instead of working with other models. Also I made this meme with Hermes and it was able to switch through many models to get good video output!#fable5
I very often tell amongst my friend circle that we might be the last of the generations that actually die. I m incredibly bullish on the latest research in personalized meds and the effects of AI on our future health.
This has quietly been a miracle month in medicine.
In the last 5 weeks we’ve got news on:
- retatrutide, the triple agonist GLP-1 from Lilly, basically melting fat and body-wide inflammation at record levels
- RevMed’s new pancreatic cancer drug showing unprecedented abilities to extend life
- small trial of a one-and-done PCSK9 gene editing therapy for slashing LDL cholesterol
- Mayo’s AI-assisted radiology showing vastly improved cancer detection
- this new therapy for metastatic solid tumors
This stuff is at varying levels of evidence. Retatrutide is ~100% on its way, other stuff needs more clinical trial data. But put it together and we’re maybe on the verge of majorly reducing the mortality of heart disease and cancer, the two leading causes of death in America.
AI feels like it’s entering its “memory architecture” era.
Bigger models matter, but so do context windows, RAG, agent state, long-term memory, caching & knowledge graphs.
Similar to PCs/servers: CPUs mattered, then caches, buses, NUMA, and distributed systems changed the game.
I like blockchain tech quite a bit because it extends open source to open source+state, a genuine/exciting innovation in computing paradigms. I'm just sad and struggle to get over it coming packaged with so much braindead bs (get rich quick pumps/dumps/scams/spams/memes etc.). Ew
The KelpDAO exploit is the scenario we built @glacislabs to prevent.
1/1 DVN = single bridge exposure. It's the opposite of redundancy, and 47% of OApps are still configured this way.
XERC20 is a great answer. Glacis Core is another. Even within LayerZero, safer configs exist, they just don't have BD teams outspending war chests to push them.
Security gets de-prioritized until it's a forcing function. $292M later, here we are. Build for resiliency.
Finally a worth heavyweight to challenge Openclaw. It’s not just about better answers it is about getting better with every answer. Excited to giving it a try especially with KIMI2.6 and Claude orchestration setup.
Meet Hermes Agent, the open source agent that grows with you.
Hermes Agent remembers what it learns and gets more capable over time, with a multi-level memory system and persistent dedicated machine access.
We built @glacislabs to address exactly the issue of single bridge hacks. People in crypto cared less about security than they talk about security. XERC20 is great but doesn’t have fancy BD teams selling vendor lock-ins. Solutions exist even within L0 for better security models.
Personal experience: 4.7 uses meaningfully less tokens for same task than 4.6. It gets to answers quickly and runs in background longer with less tokens used. While reinstalling it found an old 1tb ssd that ‘failed’ 3 years ago and fixed ssid issue and it got it working !