Alpha caller, MOD, CM & Discord Dev. open for hire, I can manage your project from ground level active 24/7 dc: samahash tg: samahash112
unlimited void
- Tweets 17,272
- Following 2,955
- Followers 3,848
- Likes 20,726
Before you build a bin of stocks, build a group of people.
I'm leading Unstoppable in the Poise campaign by @PoiseFinance handle. 10 slots.
Big accounts, small accounts, all conviction.
#WhosInYourGroup
Join through this link: poise.finance/campaign/g/uns…
Introducing Poise Engine.
Turn your thesis into an index, and the index becomes a tradable asset on @solana.
Backed onchain. Priced at NAV. Pairable on day one.
Limited access: Poise.finance/campaign
14
13
237
DΞ'Sama H retweeted
Introducing Poise Engine.
Turn your thesis into an index, and the index becomes a tradable asset on @solana.
Backed onchain. Priced at NAV. Pairable on day one.
Limited access: Poise.finance/campaign
Made with AI
513
661
957
93,349
why Pact chose Aptos.
when your typical loan is only $50, blockchain fees matter a lot more than they might seem.
Pact processes thousands of loans, so paying even a few cents for every transaction can quickly work against the entire model.
that’s where Aptos comes in.
low transaction costs, massive throughput, and the reliability needed for lenders processing up to 1M+ transactions a day.
cost plus speed.
Interview w/ @aptAlix
- @pactfinance
- @Aptos
50
41
239
Hunch
I can’t believe @pp_privatejet_2 is a lady 😱😳😳😳
$50k given out already
1
12
151
watch the stream live on Kick
- kick.com/pp-privatejet
48
DΞ'Sama H retweeted
the bank used to be the middleman.
what happens when the middleman becomes code?
structured notes have existed for years.
but bringing them on-chain isn't as simple as putting a traditional product into a smart contract.
you need both sides of the trade.
that’s where @notesystems robinhood:0xc4f730335fb9e439ca5552f7b52b8e638c4245b0 gets interesting.
🧵 + Demo video 🎥 🔻
112
2
93
1,531
all design done using canva , took about 15mins...
KiiChain is building something crypto doesn't talk about enough: better FX infrastructure.
Most onchain products are built around assets.
But before you can move, trade or settle those assets across borders, you still have to deal with currencies.
That's where @KiiChainio gets interesting.
here's a breakdown 🧵 🔻
1
11
198
KiiChain is building something crypto doesn't talk about enough: better FX infrastructure.
Most onchain products are built around assets.
But before you can move, trade or settle those assets across borders, you still have to deal with currencies.
That's where @KiiChainio gets interesting.
here's a breakdown 🧵 🔻
33
1
43
1,435
06 — The bigger picture
The interesting question isn't:
"what feature did KiiChain launch?"
It's:
how many pieces of the FX stack can they actually put onchain?
Because if FX becomes native infrastructure rather than an external service, it opens up a different way of thinking about cross-border financial applications.
1
2
17
07 — That's the part I'm watching.
KiiChain isn't trying to make FX sound futuristic.
They're gradually turning the infrastructure into something users can actually interact with.
And the releases are what make the thesis worth watching.
Explore what @KiiChainio has shipped:
link: pay.kiichain.io
Then watch what gets added next.
1
22
most robot datasets are basically a photo album.
- same kitchen.
- same objects.
- same lighting.
you train on it, then someone moves one cup and the policy falls apart.
@axisrobotics takes a different approach.
instead of shipping a fixed dataset, it gives you a task generator.
prompt a scene, define the objects and goal, then the engine generates variations across layouts, assets, physics, and camera conditions.
each run is a cousin of the last one — not a clone.
that’s the difference between repeating data and actually expanding coverage.
1
1
92
1/ a static dataset has a ceiling.
you can collect 50k demos of “pick the mug.”
but you still mostly taught the robot: mug + that table + those conditions.
generalization needs controlled variation across object, pose, lighting, embodiment, language, and more.
that’s difficult to sample intentionally at scale with a traditional dataset.
2/ @axisrobotics treats diversity as part of the product.
the task generation engine breaks a goal down into:
scene → objects → behavior → success condition.
then it generates variations across layout, assets, visuals, and robot embodiment.
40 layouts × 4 assets = 160 versions of the same skill.
that’s the point.
3/ every task family comes with a checker.
if a task can’t be scored automatically, it doesn’t belong in a scalable teleoperation loop.
generation without evaluation is just more video.
4/ this creates a flywheel.
training reveals where the model fails.
those failures become signals for the next task families.
the engine generates the data the model is missing.
1
31