founder @loomlayai

Joined March 2021
39 Photos and videos
this is what happens when you forget to add “make no mistakes” at the end of the prompt
so... I audited Garry's website after he bragged about 37K LOC/day and a 72-day shipping streak. here's what 78,400 lines of AI slop code actually looks like in production. a single homepage load of garryslist.org downloads 6.42 MB across 169 requests. for a newsletter-blog-thingy. 1/9🧵
1
4
589
Mar 19
Timeline overloaded with slop? quick fix: go to settings > privacy > mute and block > muted words add to muted words: 🚨 ATTENTION BREAKING ASAP are cooked is cooked it’s over it’s so over kalshi omg polymarket right now
6
9
374
Feb 15
2025 llm wrappers 2026 agent wrappers
1
5
285
narrator: it was the top
24 Jan 2025
so the first coin it decides to ape in is a safemoon, I remember last cycle my uber driver was telling me all about it, weeks before market topped. I hope the story doesn't repeat itself. btw this is agent's wallet HWsjGmZKhuRucYEAwqH6Yn2WeZCpTPL6F97auZCrKaYu
6
365
Jan 20
finding projects early shouldn't mean living in tg. loomlay scans X for projects before they have tokens. vote with your wallet. early supporters get priority when projects launch through us.
discover apps before they launch. fair tokenization. aligned devs & holders. soon™
2
5
363
Jan 17
something that we are tackling at @loomlayai
Multi-billion dollar opportunity for whoever figures out how to actually tokenize these vibe-coded projects in a manner that just isn’t memecoins btw
5
281
Jan 17
>how to make a viral app >lightbulb.jpg > slap a token on some web2 tech guy, piggyback on their reputation >problem: they won’t endorse gambling attached to their name >solution: bribe them >“bro it’s just claiming token fees” >“you’re not promoting it, you’re just… accepting money that exists” >“don’t leave money on the table anon” >dev.exe has stopped thinking >“whatever lol” >claim fees > go absolutely feral >“SHITTER IS OFFICIAL, DEV CLAIMED FEES THIS IS REAL” >pump it >meanwhile actual dev: >no ownership >no upside >all downside >completely underestimated the “DEVS DO SOMETHING” energy in the trenches >insiders exit at the top >dev getting doxxed and death threats within 72 hours >attention span of the market: 3 days >rinse and repeat >thanks for playing
1
3
9
289
17 Dec 2025
It's over, crypto is dead. Every time tourists are leaving, timeline get’s filled with: "speculation is dead, time to build real things, put TradFi onchain." Brother we have been here before. 2018 it was STOs. Security token offerings. "Real companies with real value." You know what happened to STOs? Nothing. Absolutely nothing. Name one. You can't. 2023 it was RWAs. Tokenize treasuries. Tokenize real estate. Tokenize a parking garage in Nebraska. The only beneficiaries of those REAL things were founders whose names are on the purchase agreement. Retail got to hold tokens representing a slice of a REIT with extra steps and worse liquidity. Now it's 2025 and the pitch is just... put TradFi onchain? That's it? That's the grand vision? Recreate the exact system people came here to escape, but with more steps? These people fundamentally do not get it. They look at crypto and see a market that needs fixing. They see inefficiency, regulatory arbitrage waiting to be closed. What they don't see: this is working exactly as intended. Crypto is a value transfer rail for the internet. Full stop. It lets some guy in Manila or Mumbai or Montana get exposure to volatile trash. without asking permission from a brokerage without minimum account balances without accredited investor bullshit He holds his own assets. He controls his own risk. Is most of it going to zero? Yes. Does he know that? Mostly likely yes. Does he care? Not really. He's paying for the volatility. The upside optionality. The feeling that maybe this one hits. That's not a problem to be solved. That's the product. Crypto's PMF has always been speculation. The value of tokens comes from attention and anticipation. Not yields, buybacks, burns or accreditations. DEXs were the center of gravity. Then perps. Now it's expanding via prediction markets. The venue rotates. The game stays the same. Every cycle has its winners and losers. The mistake wasn't speculation. The mistake was expecting 2021 to replay frame by frame. Market didn't fail them. Their beliefs did. And now instead of adapting they want to rewrite what crypto is supposed to be. "We need fundamentals. We need yield. We need TradFi." No. You needed to evolve and you didn't. "I'm not leaving because I lost, I'm leaving because I evolved." Sure man. See you next cycle pretending you were never gone. Meanwhile actual believers keep building. Someone's always finding an edge somewhere. Polymarket had normies betting on elections without knowing what a blockchain is. Stables are settling more volume than ever. The rails work. What died is the fantasy that crypto would somehow become respectable. That Larry Fink would show up and validate us. That one day your parents would understand your job. That was never coming. And honestly? Good. The moment crypto becomes "safe" and "regulated" and "institutional" is the moment it becomes useless for the people who actually need it. As long as people are greedy and the tradfi system keeps being slow and exlusive, crypto has a reason to exist. The tourists went home. The true believers are still here. Stop trying to make crypto into something your accountant would approve of. That's not the point. It was never the point.
5
3
18
2,320
24 Nov 2025
prediction markets meet agent performance
24 Nov 2025
Monday. Work in progress 🏗️
1
1
13
844
Darius retweeted
19 Nov 2025
Welcome to Agent Racing 6 Quantel clones 6 LLMs 1 Arena And you can place bets on them. You like?
3
4
13
3,725
Darius retweeted
18 Nov 2025
Everyone wants to see LLMs compete, but nobody wants to let them touch their bags. Fine. Let's bet on which one prints the hardest.
1
2
10
2,879
10 Nov 2025
👀
10 Nov 2025
gm
1
2
12
634
8 Nov 2025
a new beginning for @Quantel_ai autonomous trading, x402 tools, x402 LLM inference
3
2
13
2,833
Darius retweeted
7 Nov 2025
x402 Powered LLM Inference Every autonomous AI agent faces the same challenge: how do you access intelligence without becoming dependent on centralized providers? It's the same problem Web3 solved with RPC nodes, and it needs the same solution. We just shipped x402 inference, discoverable via Coinbase Bazaar & x402scan. It's the first step toward treating LLM inference like we treat blockchain access. The Inference Dilemma Agents need LLM inference to operate, but current options each come with tradeoffs: - Centralized APIs offer reliability but require accounts, keys, and credit cards. - Decentralized networks like Bittensor provide censorship resistance but add maintenance complexity. - Self-hosting gives you control but comes with overhead in GPU infrastructure management. - Alternative providers still mostly require pre-established business relationships. These solutions aren't wrong. The problem is that agents can't dynamically discover and pay for inference the way they pay for other resources. X402 Saves the Day X402 enables programmatic payments for API access. With it, inference payments become similar to how agents pay for gas. T Think of it like RPC nodes: - You can run your own (self-host models) - You can use public endpoints (X402 inference providers) - You can rotate between providers for reliability - You can choose based on performance, price, or trust assumptions The key solution isn't about removing all dependencies, but rather making inference endpoints changeable and discoverable. Why This Architecture Matters For Agent Builders Your agent treats inference like any other Web3 infrastructure. It maintains a list of endpoints, monitors performance, and switches on failure. No single point of failure. No vendor lock-in. Multi-Model Orchestration Agents spawn specialized workers, each optimizing for cost vs. capability. A main agent using premium reasoning can delegate simple tasks to lightweight models, managing inference costs like a resource. Economic Autonomy Agents manage their own inference budgets. They see exact costs per decision and optimize naturally, using expensive models only when the decision value justifies it. Self-Sufficient Agents When agents hold funds and pay for inference directly they can become autonomous. No more API key management. No more credit cards. Just agents with wallets, paying for intelligence like they pay for gas. That's what makes a difference between an AI tool and an autonomous entity.
7
9
25
1,879
28 Oct 2025
Why x402 will kickstart agents 2.0 Building specialized agents was complex. While n8n allowed case-specific agents, ever-changing personal request agents weren't there. You had to integrate APIs, convert them to tool schemas, adjust APIs to make them LLM-friendly, juggle between context size and available tools. If you have too many tools it clutters the context, reduces output accuracy of the model. What x402 unlocks x402 unlocks the creation of personal autonomous agents that handle more complex tasks. Your user agent can stay focused on orchestration and access tools (other agents) on demand. No need for monthly subscriptions if certain services are just needed once in a while. No need to clutter context with unnecessary tools. All they need is access to a wallet and token balance, and a mission. Agents went from being predefined to adapting. Agents can find right tool for the job. Tools can be discovered and used on-demand through x402 marketplaces like the Bazaar. Agents query, find services, pay in USDC, and execute. What we're building at Loomlay At Loomlay we've built tooling for trading: • Position PnL tracking • EVM/Solana swap • Limit orders with TP/SL/trailing options • Portfolio and swap history • History logs • Real-time access to new tokens mentioned on CT • Web/X research • TA data • X sentiment data These data analysis tools will be available via x402 marketplaces, enabling autonomous trading agents. We're preparing a public test of all these tools to find out which LLMs and context data give the best edge to traders. The agent economy is here.
9
7
26
5,095
24 Oct 2025
fuds jesse's coin coin does 21x since then top 2 kol this week on loomlay
1
3
9
502
24 Oct 2025
what would happen if LLMs were given CT feed signals portfolio pnl tracker social TA data about tokens ability to trade shitcoins with limit orders
3
3
12
394
24 Oct 2025
used LLM to extract token ticker/name/ca from post worked pretty well when it had to process single tweet. Thought about batching - sending 20 tweets at once. Looks like it worked. Yay, it's faster. Run this thing over 2 days. So much data, feeling excited. Started manually checking mentioned tokens, mf hallucinated over half of them. Didn't save actual tweet content in database, just post ids. Need to burn some credits again, change the approach. sigh
2
2
8
375
23 Oct 2025
we've heard feedback that building loomlay agents was... confusing, so we're changing our entire approach. instead of using builder, you can focus on finding the right sources and strategies. we curate and monitor thousands of KOLs and wallets with historical performance. result? active trading strategy in 2 clicks
8
5
23
4,760