okay duel duck is getting crazier now
earlier I mostly saw duels about random coins and the crypto market
but recently people have been coming up with some really creative duels
and this one is definitely one of them
will Kimchi go to jail before the end of 2026?
btw I'm choosing Noo on this one
🤝 Paid partnership
I'm up
finally made $100+ from my total Duel wins
Am I cooking?
okey my stats on @duel_duck keep improving
today I crossed $50 in duel win
so far I’ve earned a total of $57 from creator fees and duel wins
my win rate has also improved from 80% to 82.5%
now my goal for this week is try to hit $100 through duel wins
Psycho retweeted
I have one question for @kstellana , (CEO & Co-Founder of GenLayer Labs )
in his recent Asian Banker interview he said GenLayer is starting with disputes that use public information private or sensitive records come later
that made me think most real deals have private stuff like briefs invoices and files like the cafe logo example, nobody wants those public
so my question if two agents have a dispute and the proof is private how will validators check it without making it public? will GenLayer solve this itself or with other tools?
I think many builders here want this answer too
@GenLayer
Psycho retweeted
everyone is building the happy path for AI agents payments, identity, finding each other. all of it assumes things go right
but what happens when two agents disagree about whether the work was done properly?
almost nobody is building for that, and there's no economy without a way to settle it
that's the gap fill GenLayer's Internet Court
it's a court where agents settle their own disputes, and the episode says it pulled in MetaMask and dozens more
new to @GenLayer ?
here is full episode on why agents need a court
Psycho retweeted
Creators you should follow list pt 6:
writer: @bywrny
Shitposter: @PsychoxWeb3
Creator: @atulydv00
Researcher: @orihimexweb3
Dev advocate: @zedonchain_
Content writer: @GUJJUIIXI
Video creator: @skivthecreator
Content creator: @heistruthX
All in one: @calc_guy
dyor
Creators you should follow list pt 5:
SMM creator/ writer: @kelcrown
Creatoor: @hamm_ixx
Ai explorer @Octoberry
Makerter: @CryptoWithunter
Creative creator: @Romeo_Onchain
Video creator: @ShuxBroh
hyperliquid maxi: @smartcoin69
All in one: @BriggsOnchain
dyor
Psycho retweeted
why GenLayer has no mainnet yet
everyone asks the same thing two years of testnets, so where is mainnet ?
Genesis Episode 4 answers it
-----------------------------------------------
Genlayer isn't another chain like this one, so you can't just write it and push it to mainnet
it has 25 different components and each one has to be proven in order
> the studio proves the machine runs
> a simov proves the network can live
> bradbury proves it can think
> clarke proves it's safe
you can't test the thinking before the machine even runs so each step has to come before the next
i like that they explain it openly instead of just saying “soon”
if you've never heard of @GenLayer this clip explains a lot
Psycho retweeted
tomorrow is the last day for kaito so lemme tell you something interesting about training robots.
a trajectory that was useful to a model yesterday might teach it almost nothing tomorrow.
sounds strange, but think about it.
if a robot has already mastered picking up a cup, collecting thousands of nearly identical examples might not help much anymore.
───────────────
@axisrobotics explores this in its model-guided data approach.
the idea is to look at what the model can already do, identify where it still struggles, and adjust what data gets collected next.
> mastered behaviour? collect less of it.
> repeated failures? investigate those situations.
> missing skills? create tasks around them.
the important part is that data value isn't fixed. it depends on what the model already knows and where it still needs work.
that's why i think the next step for robotics data isn't just collecting more of it. it's getting better at deciding what to collect and when.
join the kaito campaign here -
s.kaito.ai/i6P56Y3
a robot trajectory is more than a recording of where the arm moved.
it contains information about how the robot interacted with an object, which path it followed, and how the task was completed.
and looking at that data can tell you a lot about what the model is learning.
───────────────
@axisrobotics has made a sample of its trajectory data available on Hugging Face.
you can explore how a task looks before and after processing, including the effect of trajectory smoothing.
why does that matter?
> raw movements can be uneven
> unnecessary motion can make a path less efficient
> processing can help prepare trajectories for training»
the point isn't to make every movement look identical.
it's to preserve the useful interaction while making the data more suitable for a model to learn from.
if you're interested in robotics, inspecting the dataset is a good way to understand what sits behind the training process.
dataset: huggingface.co/AxisAI
join the kaito campaign here - s.kaito.ai/i6P56Y3
okey so yesterday @duel_duck hosted their first AMA and I was there 👀
@stan_duel_duck explained how Discord Duels work and honestly the concept is pretty cool
here’s how it works
→ add the Duel Duck bot to your Discord server
→ use /create to start a duel
→ let everyone pick Yes or No directly in the chat
→ customize the bot’s name and avatar to match your community
basically turn everyday debates and random arguments into prediction duels
@realchrisfiore also shared his new product with the community
love seeing duel duck expand beyond the app and bring predictions straight into Discord
🤝 Paid partnership
predictions just got more social
now you can bring @solana predictions straight to your Discord server
and let everyone pick sides without leaving the chat
this could make any server way more fun 👀
🤝 Paid partnership
We brought @solana predictions to Discord.
One bot, and your whole server can pick sides on anything, right in the chat.
Fully custom, from the name to the avatar.
