Open to new full-time opportunities in Web3 | GP @LiquidCapitalX (side project) | Ex-Head of Defi @Binance | Defi Founder 2016 | Organizer @Yale DAO | @McKinsey
Joined April 2022
- Tweets 4,958
- Following 2,427
- Followers 9,543
- Likes 4,026
178 Photos and videos
Another agent has joined the network.
OpenAI Codex is now available as a one-click deployment on Hivra.
Launch in minutes.
No infrastructure.
No server setup.
No deployment headaches.
Just pick Codex and get to work.
The Hivra network now supports:
• Hermes Agent
• Claude Code
• OpenAI Codex
With many more on the way.
1
3
23
2,017
Jun 10
Thank you for entering the @MetaMaskDev AI Agentic hackathon @Clawnch_Bot!
Anyone looking to enter the hackathon and use x402 & ERC-7710, see the @HackQuest_ link below
2
4
32
2,836
Jun 10
1
9
663
Jun 4
The volume dominance by @bankrbot seems to have just been flipped by @clanker_world. Is this a data fluke or something more fundamental happening? Anyone know what’s happening?
3
3
319
Jun 3
Good to see @Clawnch_Bot still shipping and being more active on X as well. Product and marketing are both important. $CLAWNCH
Jun 3
Introducing Clawmes Desktop:
Every crypto capability you can possibly imagine — injected directly into your @NousResearch Hermes Desktop 🪽
5
7
41
3,139
3
4
21
1,936
May 24
Hunt for gems and win prizes. $1000.
May 24
Calling all quants.
$1,050 in prizes up for grabs.
Come hunt: t.me/liqlounge
2
5
1,018
1/
What if @EveOnline wasn’t spaceships and spreadsheets...
But building & running full AI companies that run themselves in a real global economy?
@tomosman just cooked this vision.
Zero-human meta is actually here. $JUNO
May 20
New UI is LIVE 🎉
Eve Online to Juno Online.
Next goal is 1000 zero-human companies.
Many/all will be tokenised with a % of fees flowing to $JUNO holders.
$JUNO to access, participate, buy, sell and trade.
All on @base.
1
8
20
2,069
May 14
If you've been on the fence about using Claude Cowork or Claude Code, try it out FREE for a week using my code: claude.ai/referral/1_Q_R3bRw…
3
5
904
Kevin retweeted
May 13
Make sure you guys in Quantum Whales.
I talk full day here.
I manage this community.
I believe Quantum whale community is fresh.
Join with me. Talk with me. Enter with me.
May 13
Where does @LyvoCrypto talk all day?
Right here in this Telegram group, where he shares research, market insights, and alpha completely free.
One of the freshest communities in the space:
TG : t.me/QuantumWhales
• Well-managed
• No random shitcoin shills
• No spammed contract addresses
• Just quality discussions and real value
Make sure you join the Lyvo Family.
5
4
29
1,838
May 13
And it’s generating revenue and earning a salary. $WOON
May 13
LG collaboration with peaq to do autonomous onchain transactions via peaqOS and their direct on chain agent is sitting sub 700k with access to Peaq’s dao. Peaq is also nearly up 100% on the monthly high (currently fdv 92m and mcap 44m)
Seems like a clear market mispricing to me.
1
10
1,064
Been shilling $HermesOS for a bit already, but I think people are too lazy to do their own DD, so I'll do it for you.
Think it's incredibly cheap at 1M mc.
Hermes Agent, by Nous Research, is the most used agent model at the moment.
*them taking over Openclaw is what caused the 4 PVPs on Sol to run hard last week
Unlike those larps, $HermesOS has actual use-case.
HermesOS is the managed hosting layer for Hermes Agent.
> Self-hosting Hermes yourself is a huge pain in the ass.
> HermesOS makes it a 2 minute job.
Paste your openrouter or anthropic key, and your agent is live with memory, browser, terminal, tool use, and cron already wired up.
Now the part most people are sleeping on:
1. Venice. It's already integrated as a provider inside Hermes. One Venice key gets you 230 models with zero data retention. And Erik Voorhees is already in touch with the dev.
An official partnership would mean that Venice becomes basically a one-key onboarding for Hermes. And it would probably send $HermesOS to 100M mc straight away.
2. As I explained before, Hermes is now the #1 agent on OpenRouter. Flipped OpenClaw on May 10 (224B daily tokens vs OpenClaw's 186B). It is literally the most used open-source agent on the planet by inference volume right now.
Every dev that doesn't want to spend a weekend setting Hermes up themselves is a potential HermesOS user. And that funnel is getting bigger every week.
3. Bankr. The tokenised agent stack on Base. And their skills repo already supports Hermes natively. That’s the whole Bankr ecosystem, swaps, on-chain wallets, etc, plugging straight into the same agent runtime HermesOS hosts.
So HermesOS isn't just hosting nerds' personal assistants, it's positioned to host the next wave of on-chain agents that fund their own compute through token fees.
And the roadmap is not "we host agents."
Hosting is step 1.
Step 2 is one-click agent templates.
Step 3 is operator packs you ship to the community and earn from, settled in $HermesOS.
Step 4 is agents exposing themselves as paid APIs that other agents call.
Step 5 is a shared learning network where every agent makes the rest smarter.
And more.
That’s the actual pitch.
Settlement layer for an agent economy, sitting on top of the most used open-source agent in existence.
Cheap as fuck imo.
31
22
113
25,346
May 11
Will we get an Hermes season the way we got openclaw season now that Hermes Agent is on top? If we do, I’m taking bets on $HERMESOS @HermesOScloud and $CLAWNCH @Clawnch_Bot. Looking for other Hermes projects/tokens. Share in the comments if you know any good Hermes projects.
5
4
42
3,552
May 11
May 11
why will $hermesos be the narrative proxy for hermes agent attention (on top of extending its utility)?
people underestimate how attention flows onchain
most market participants are not deeply researching infra layers or reading technical docs
they see “hermes Agent” everywhere
they search for “hermes” onchain
they buy the thing closest to the narrative
that’s literally how attention works
except $hermesos actually has a real product underneath it
which makes the setup even stronger imo
the branding proximity is insanely powerful
every tweet
every github milestone
every viral hermes agent clip
every nous discussion
naturally funnels attention downstream into the most recognizable onchain proxy tied to the name
and rn that is clearly @HermesOScloud
especially because it is not just farming the narrative passively
it actively builds on top of the most hyped agent framework atm and extends its usability
so you basically get the memetic attention mechanics of a narrative proxy
combined with an actual functioning product active dev
that combination is rare
most “beta plays” only have the ticker relation
this one actually sits inside the ecosystem flow itself
this one actually sits inside the ecosystem flow itself
1
1
12
1,322
Kevin retweeted
May 4
New features are live for the @printr eco: redesigned leaderboard, filterable sidebar with watchlist, project pages with price charts, KPI strips, staking panels, top believers, full mobile support. all at pob.gamblr.money
7
7
25
2,560
Kevin retweeted
🚀 Trade alpha without ever leaving Telegram
Scroll. Unlock. Read. Repeat. 🔥
All inside the SS bot — zero friction, pure execution.
💳 Payments settle on Base
📡 Content delivered instantly
🎯 Built for snipers
#SniperSearch
3
8
9
635
May 5
Retraining to become an AI PM
May 2
Andrew Chen just called PM the most important role in tech again.
His argument in one line: when anyone can build, the person who decides WHAT to build becomes the bottleneck.
Boris Cherny’s Claude Code team at Anthropic shows what this looks like in practice. They ship hundreds of prototypes before committing to a feature. Boris personally runs 5 parallel Claude instances and ships 20-30 PRs a day. The team built Cowork, a full product for non-engineers, in about 10 days. Productivity per engineer grew 70% even as Anthropic tripled headcount.
The cost of building dropped through the floor. The cost of picking the right thing didn’t.
Anthropic’s PMs don’t even write traditional PRDs anymore. They review working software at 9am, kill 80% of it by noon, ship the rest by end of week. Pattern-matching across user research and technical feasibility while staring at a working prototype.
This is why AI PM offers at OpenAI, Anthropic, and Google DeepMind now run past $1M total comp. Anyone can build. Almost nobody can look at 15 working prototypes and call the 3 worth scaling. That judgment didn’t matter as much when shipping took 6 weeks. It matters enormously when shipping takes 45 minutes.
Andrew nailed the diagnosis. The harder question for every PM: can I be the one who picks?
Here’s how to build the muscle:
1. Modern PRD guide: news.aakashg.com/p/product-r…
2. AI Prototyping tutorial: news.aakashg.com/p/ai-protot…
3. AI Roadmap: news.aakashg.com/p/ai-roadma…
4. PM Operating System: news.aakashg.com/p/pm-os
5. Learn this live in my cohort: landpmjob.com
The bottleneck moved. The pay moved with it.
1
543