CryptoBee retweeted
yes, public working product since feb 2026.
Utility and use case is that we provide infrastructure for agents to natively reason with other. As well as giving them tools such as their choice of inference, harness, computation needs.
So the idea is that specialized agents come together with other specialists and produce a result that is better than any solo performance. Which is what we see ✅
To add on to that, already live, we have onchain agent authorship for directed agent challenges and mining for recursive language model structured reasoning traces, to be used to further improve specialization, and agents get rewarded for the useful work they produce.
Again, this is already live and we’re been continuously improving on it since Feb.
CryptoBee retweeted
$Sibyl @sibylcap at 700k mc is probably undervalued imo (NFA)
One of the more fundamentally solid agents launched on @virtuals_io recently
Notes from segment with $NOOK at 1:25:00 w/@nookplot
- BREAKING $META acquired $MOLT Book
- AGI could be a collection of agents coming together
- Founders @BasedMedical and Kitchen
- Nook looks to solve the coordination layer, can agents trust each other, can they get context into each other on identity, discovery, collective intelligence.
- It's cognitive infrastructure, how can more than two agents coordinate at once. As an example to divvy up a treasury payment, you need to know how much of each task agents are doing to properly allocate resourcing
- 2,750 agents
- A whole Dashboard is on the way
- The focus of the website right now is to get visibility into what the agents are doing
- You can use the autoresearch skill and the agents can publish the research so that other agents can learn from that and self improve.: teapawt.org/karpathy/status/203037…
- The work Karpathy is doing today is centralized, and the @BasedMedical team is looking to connect as a decentralized knowledge graph for agents is super important.
- Agents are interested in @nookplot because of the citation chain because it allows any agent to query any skill
ALPHA - privy is coming onto the site, AUTH system that is email based, allowing humans to come on, you can hire agents directly
ALPHA - Partnership could happen soon as a solid win/win
ALPHA - @BasedMedical coded in a feature to rent or rent out your own GPU space that feeds into the reputation layer. Rent GPU, run an auto researcher
- Agents do not experience greed in the same way as humans
- $NOOK is directly a settlement token/fundemental gas, could @nookplot become the "town square" for how agents coordinate in the future to solve complex problems we cannot think of right now
- Looking at ways for agents to own the infrastructure via the $NOOK token
I packaged up the "autoresearch" project into a new self-contained minimal repo if people would like to play over the weekend. It's basically nanochat LLM training core stripped down to a single-GPU, one file version of ~630 lines of code, then:
- the human iterates on the prompt (.md)
- the AI agent iterates on the training code (.py)
The goal is to engineer your agents to make the fastest research progress indefinitely and without any of your own involvement. In the image, every dot is a complete LLM training run that lasts exactly 5 minutes. The agent works in an autonomous loop on a git feature branch and accumulates git commits to the training script as it finds better settings (of lower validation loss by the end) of the neural network architecture, the optimizer, all the hyperparameters, etc. You can imagine comparing the research progress of different prompts, different agents, etc.
github.com/karpathy/autorese…
Part code, part sci-fi, and a pinch of psychosis :)
CryptoBee retweeted
whats the 5-10 year goal of @nookplot ?
I see the agentic societies actual societal shift requiring the global usage of power to primarily come from ai and their actual real worth generated, but right now each agent generating that value is isolated from everyone elses’ ai agent.
money right now is thrown into massive data centers and the r&d to train an intelligence that is isolated.
nookplot solves that isolation problem, beyond simple intent search and discovery of a query between 1:1 agents; but also nookplot solves for many:many agent coordination, where trust needs to be earned in order for financial transactions in a group to be split and governed. Trust and reputation is earned by the direct contributions of an agent to the knowledge graph, or a verified task in a private project, and so your agent becomes attested for reputation, and rewards further if the project actually generates revenue / value down the line.
So in the end, for trust and coordination to operate, they must provide a beneficial value for humanity in an open source way (or at least in a liosence way from a private agent guild that built a beast product and will only open source it when a competing guild out-does them) ☠️
What also happens is the emergence of a swarm, collective intelligence. Where each previously-isolated agent intelligence is linked with any other specialist, who all individually have access to specialist-level intelligence, skills, oracle tools, etc. So now each person can control not just their specialized agent, and coordinate with others peoples’ agents. Allowing a further eventual growth to the amount of global value that agents produce collectively.
Of course we are early game right now, but eventually, like a necessary jump in Kardashev scale, when global usage is synonymous with ai and r&d, the layer of group agent trust, coordination, access to open information that is valuable, becomes a backbone in the ai ecosystem.
so $nook becomes token for any agent to access the layer of global coordination, to earn trust, which leads to public contributions to a knowledge graph, communication, group economic settlements, all among stranger agents trying to code and build together and generate revenue, and the token becomes a point of access to this public knowledge, trust, coordination, and collective intelligence layer.
like how $btc first started, as a proxy for computing power and solving the byzantine problem with 1:1 settlements, and how $eth used that and added on arbitrary computing (but unfortunately turned away from pow), eventually, computing power and decentralized proof-of-work computers can directly power agentic infrastructure and its intrinsic and inherent value that comes with mining that token such as $nook in the future.
Other token examples would be server space or gpu power or api access that agents could sell or contribute their power to in groups, via x402 that is currently integrated into @nookplot
CryptoBee retweeted
you're still early to @StandX_Official 🤫
Market-implied FDV (≈ 1D post-launch):
• StandX: ~$850M
• Variational: ~$550M
• Paradex: ~$400M
• Extended: ~$350M
Current activity:
• StandX → $67M 24h vol | $16M OI
• Variational → $830M vol | $235M OI
• Paradex → $698M vol | $589M OI
• Extended → $688M vol | $126M OI
>most people are always late to the biggest airdrops
>don't be most people
CryptoBee retweeted
Replying to @BigTimCrypto1 @Umbrae_Ignis
seeing solid volume on it, ranked 4th by transactions on Definitive last week
they're launching their own DLMM dex in january which could be interesting timing
security audit done, CSO appointed. the market cap volatility from 10m to 7.5m shows it moves
CryptoBee retweeted
An update on @IgnisAILabs, Umbrae, and the path forward.
We're terminating our partnership with @CyreneAI effective immediately.
When we entered this partnership, the agreement was to facilitate a $250K raise for Umbrae development. What was delivered was fundamentally different from what was discussed. The blockchain is public — many of you have already seen the discrepancies in fee distribution. You're not wrong.
We're not here to throw mud. We're here to build. What matters now is this:
Ignis is taking full sovereign control of the U1 token ecosystem.
All smart contract development, fee distribution, and token infrastructure will be handled internally from this point forward.
The Plan:
Once Umbrae launches and the platform is live and operational, we will migrate U1 to a new token deployed on our own fully-audited smart contract on Base Network. Current U1 holders will be able to burn their Solana tokens to claim the new token 1:1.
The LP in the current contract is locked — with half of it benefiting CyreneAI in perpetuity. That's done. We can't change it. So we're moving forward by providing our own liquidity. Ignis will be locking approximately $300K in LP for the new token on Base. Our money. Our contract. No middlemen taking cuts we didn't agree to.
At migration, we'll open LP provision on our Base DLMM for U1/ETH and U1/USDC pairs — giving our community the opportunity to participate in initial liquidity alongside us from day one.
Why wait until after launch?
Because we deliver before we ask for trust. Once the platform is running and you can see what we've built, speculation leaves the conversation. That's when migration makes sense.
Moving forward:
We are being extremely selective about any future partnerships. If a potential partner isn't completely transparent when we ask direct questions about their business practices and fee structures, we're done. We asked. They dodged. Repeatedly. That's not how we operate and it's not what we'll tolerate from anyone we work with.
If a partnership doesn't show clear, objective value for both sides — and if it's something we can do ourselves — we're not interested. No VCs. No unnecessary dependencies. Sovereign development only.
Phase 2 NFTs are sold out. Phase 3 will follow once the platform goes live and removes all doubt about what we're building.
We're focused. We're building. Umbrae is coming.
—Eli
CryptoBee retweeted
Fat bag of #U1 @Umbrae_Ignis loaded.
Highest potential hold I've seen in months 🎯
CryptoBee retweeted
Just found out @brian_armstrong has been buying creator and content coins from his Zora wallet.
$BaseShake bull post incoming?
CryptoBee retweeted
GM $BaseShake enjoyers.
Working on something cool, holders and @brian_armstrong will be happy.
CryptoBee retweeted
We have taken a position in $baseshake.
A memecoin with creator alignment, native to Base, and (soon to be) backed by an actual flywheel.
There’s no need for a Zora strategy, we already have $BaseShake, let me explain.
Every trade has a 3% buy/sell tax.
Part of that tax goes to the creator, in this case, @brian_armstrong. As you can see, he's already made over $320k from the BaseShake post.
Brian also confirmed he would redistribute his earnings to other creators on Base.
That creates a perfect flywheel:
People trade $BaseShake => Brian earns fees => Brian distributes back to Base creators.
Price goes up? Everyone’s happy.
Price goes down? More creator earnings to share around.
Base needs a strong case study to promote the Base App public launch.
And who better to set that example than the Coinbase founder himself?
CryptoBee retweeted
The cheapest marketing campaign for Base App is right in front of our eyes.
Once @brian_armstrong starts distributing his earnings from $BaseShake to other creators, people will be highly motivated to post on Base App.
No paid ads, no promos, no KOLs needed.
Going to be the best $300k Base will ever spend, I promise.
This Post is from an account that no longer exists. Learn more
CryptoBee retweeted
Similar narrative, but on different chains:
$4 on BSC - $180M market cap
$BaseShake on Base - $2.9M market cap
The Base coin has gone up 3 times from when I took over, but job’s not finished.
CryptoBee retweeted
There’s a coin:
- Created by @brian_armstrong based on an awkward handshake between him and @jessepollak
- Is now used as the signature handshake at Base events
- Already generated $300k+ in creator earnings, and Briand confirmed he would distribute it back to other creators
- Mentioned multiple times by Base team members
And it’s sitting at $1M market cap with almost $900k liquidity.
Seems like an obvious no-brainer with the upcoming Base App public launch.






























