
Building @hyre_agent | Own https://xitter.cf/t.co/8tktcxtzxo | ex @dfinity | Dev @buildonorbit Privacy Tech Advocate @iEx_ec
Joined October 2010
- Tweets48.9K
- Following5K
- Followers62.2K
- Likes45.8K
Eight retweeted
Cara bikin bot trading meme Solana pakai AI Agent
Setelah gw sharing experimen trading bot yang bisa flip modal 1 jt jadi 8 juta atau 8x banyak yang nanya cara bikinnya.
Jadi gw bahas alurnya supaya kalian bisa ngarahin AI agent kalian sendiri. Belum punya ai agent? Tutorial install cek di pin post gw.
Trading meme sekarang didominasi bot. Banyak smart wallet beli lalu keluar hanya dalam hitungan menit. Kalo lu masih manual trading cari 10-20%, sering telat exit karna lawan kalian bukan manusia.
Gw nggak ngejar 100% ROI atau 10x di setiap token. Profit 10–20% aja sudah cukup asal sistemnya bisa ngulang proses ini berkali-kali.
Posisi per trade gw juga kecil tapi per hari bisa 50-80 trade. Modal besar dalam satu token rawan dibegal dev, kecuali tokennya memang punya komunitas kuat dan distribusinya sehat.
Ini workflow yang gw pakai: Diskusikan topik2 ini sama agent kalian
1. Trigger
Bot nggak perlu scan semua token. Bot cukup nunggu token dari sumber tertentu, lalu masuk ke proses filter.
Sumber trigger yang bisa dipakai:
- Dex event dari DexScreener
- Trending atau hot search GMGN
- Hot pool Meteora
- Migrasi Pump.fun
- Aktivitas kumpulan smart wallet
- Call dari channel Telegram
- Post dari akun X tertentu (API X berbayar)
Diskusikan dengan AI agent kalian sumber mana yang paling cocok.
Sebelum build, pantau dulu token-token dari trigger tersebut. Cek berapa banyak yang naik, seberapa cepat naiknya, dan kapan biasanya mulai dump.
Buat dulu tools untuk monitoring lebih enak, sebelum nanti kita tau sifat pergerakan harga token dengan filter a b c gimana
2. Filter
Setelah menemukan trigger yang sering menghasilkan peluang, bot harus milih token mana yang layak dibeli & mana yang di-skip.
Minta agent kalian ngisi semua angka filter di file config supaya gampang diubah tanpa mengutak-atik kode.
Data bisa diambil dari Solana Tracker, GMGN, Mobula, atau provider lain mungkin ada lg, pengalaman gw sndiri Mobula masih paling bagus tapi sekarang dia nggak gratis.
Bisa pakai solana tracker cuma harus ada gate jika data null, atau 0 wajib di skip (masih kurang akurat sedikit)
Contoh data yang masuk filter:
- Market cap
- Liquidity
- Volume
- Umur token
- Dev holding
- Bundler percentage
- Sniper percentage
- Top 10 holder percentage
- Perubahan harga 1 menit dan 5 menit
Tambah aja sesuai gaya trade kalian
Ini bagian yang paling lama bisa berbulan bulan.
Filter terlalu ketat bikin peluang yang masuk terlalu sedikit. Filter terlalu longgar bikin bot rajin entry ke token rugpull.
3. Security check
Sebelum entry, bot wajib memeriksa:
- Honeypot atau kemungkinan token tidak bisa dijual
- Mint authority
- Freeze authority
- Risiko tambahan dari program token yang dipakai
Gagal satu pemeriksaan, langsung skip.
4. Entry
Token yang lolos trigger, filter, dan security check baru boleh dibeli lewat API Jupiter.
Beberapa batas wajib disimpan di file config trade (jangan hardcoded):
- Maksimal jumlah posisi terbuka
- Nominal per transaksi
- Maksimal slippage
- Batas rugi & profit harian (kemaren gw pake nonstop profit sering balik ke market di jam jam tertentu)
- Prioritas fee dan batas biaya transaksi
Tujuannya supaya transaksi buruk nggak menghabiskan profit
5. Exit rule
Setelah entry, sistem harus memantau setiap posisi secara rutin. Interval 2–5 detik masih cukup masuk akal untuk beberapa posisi tanpa terlalu membebani API Jupiter.
Sistem exit yang gw pakai:
Hard stop loss
Kalau harga nyentuh batas rugi, langsung jual.
Take profit
Kalau target tercapai, ambil profit sesuai konfigurasi. Gw sendiri target 1000%, ga nutup kemungkinan ada yang langsung hard pump, tapi meski ga nyampe 1000% kita masi punya trailing stop
Trailing stop
Misalnya trailing aktif setelah profit 10% dengan jarak 5%.
Kalau harga naik sampai 50%, batas exit ikut naik jadi sekitar 45%. Pas harga berbalik nyentuh batas itu, bot nutup posisi agar profit nggak ikut hilang kebawa turun lagi.
Maximum hold time
Kalau token sudah terlalu lama dipegang tetapi harganya nggak gerak, keluar dan pindahkan modal ke peluang lain.
6. Dry run wajib
Latih bot di dry run selama mungkin.
Jangan masuk live trading sebelum strateginya terbukti profit dalam ratusan posisi, bukan hanya profit lima atau sepuluh kali.
Catat win rate, realized profit, kerugian terbesar, slippage, biaya transaksi, dan hasil berdasarkan masing-masing trigger.
Kalau dry run aja nggak profit, jangan harap hasilnya profit setelah pakai uang asli.
Disclaimer: hasil dan workflow yang gue pakai, bukan jaminan hasil untuk orang lain. Bukan financial advice. Crypto berisiko tinggi. Pakai uang yang siap hilang dan tetap DYOR.
Replying to @oezank_idc
cuman butuh how to build nya aja mas, nanti ngulik sendiri, masih bingung langkah nya, ini keman, screening nya gimana, masuk market nya gimana, keluar nya gimana, setidak nya ada clue yang bikin otak ga stuck.
Every other moat is eroding fast. What's left standing is data, and specifically, high quality data.
Once OpenAI and Anthropic launch their own software-building tools, probably within a year, the SaaS companies without real lock-in or proprietary data don't have a defense left.
Startups take the hardest hit, they had the least moat to begin with.
"Sophisticated recording device": $23.51. Antenna: $0.50. Marketing show.
xitter.cf/realMeetKevin/status/2…
Whatever new thing you're trying to build in code, chances are it's already open source somewhere on GitHub. And if it's not, someone clones it and open-sources it within days of launch anyway. The moat in coding is basically gone at this point.
Google just dropped the new Argon model, but it's Ultra only, so basically 10 people on earth actually get to use it.
xitter.cf/mattsilv/status/211004…
People keep repeating this 6 month cycle thing, but hardware physically can't move that fast. RTX Spark was announced 6 months ago and units are only just starting to ship now. The next DGX Spark will probably just bump to DDR6 or DDR7 with a small performance gain. Prices are going up, and bandwidth isn't doubling like people think.
Hear me out.
Sell your Spark clusters NOW.
Do not listen to the hype.
The things coming in the next 6 months will make that machine look like an overpriced toy.
Intel, AMD, Apple, and even smaller Chinese firms are about to dominate on price and performance. There will be no reason to even think of owning a spark at current prices. They are not good general machines either where the competition like AMD Ryzen AI Halo or Mac Studio are whole PCs that make incredible work machines in addition to providing comparable or superior inference.
Izin join vaultnya @eisbedog
kalian gak jago nge lp?? tenang, kenalin @Hedginxyz
untuk info lengkap cek video di bawah dan channel tele dibawah t.me/megumiivault
notes: nfa & dyor, beware scammer link asli cuma ini
A country declaring a cyberspace state of emergency is not a small signal. Over 20 major Japanese companies have reported breaches as attacks surge.
Cybersecurity is about to matter a lot more than people think.
xitter.cf/Polymarket/status/2109…
BREAKING: Japan declares a "state of emergency in cyberspace" as cyberattacks surge, with more than 20 major companies reporting breaches.
Readers added context they thought people might want to know
Japan did not formally declare a state of emergency in cyberspace. The phrase originated from a ruling party official at an LDP meeting; government agencies issued cybersecurity warnings and basic security advisories to companies amid breaches at 20+ firms.
japancyberwatch.com/articles/japan…
japancyberwatch.com/articles/japan…
cyber.go.jp/pdf/press/taio…
thenextweb.com/news/japan-cyb…
Honestly, no matter what field you're in, if AI agents haven't taken over at least half your daily routine by now, you're probably falling behind.
Hackers remotely defrosted refrigerators at hundreds of French supermarkets, causing massive food spoilage.
Why do these refrigerators even need to be connected to the internet in the first place?
xitter.cf/Polymarket/status/2109…
Golden Week just ended and Alibaba wastes no time. QwenGyre lands with elastic GPU scheduling and more efficient trajectory processing for long-horizon agent tasks.
This looks like the warm-up before the actual model drops.
xitter.cf/Alibaba_Qwen/status/21…
Introducing QwenGyre, bringing elastic RL to extreme-long-horizon (XLong) agent training. ✨
It launches with two core designs: 🥳
- Elastic Scheduler: reallocates GPUs between rollout & training as demand shifts—live executions never interrupted. Up to 1.85× speedup over Colocate, 1.78× over Async.
- Trajectory Processor: turns branching executions (compaction, sub-agents, retries) into prefix-shared trajectory trees with bounded, deduplicated training samples. Partial progress after timeout still gets scored.
Scaled to Qwen3.8-2.4T-A95B with ~700K tokens per rollout, QwenGyre lifts NL2RepoBench from 52.5% to 58.5% in 48 steps, cutting end-to-end training time by ~44% vs Async.
🔗 Read the paper:
arxiv.org/abs/2609.33848
I don't think running AI locally is really about saving on API costs. It's more about owning your digital sovereignty again.
Build time dropped from 6 months to a week. Reasonable, given the tools now.
What's wild is attention dropped harder, from 3 months of relevance down to 24 hours. The cycle isn't just faster, it's collapsing.
Seminggu ini mau coba platform punya @Friday_SOL
We’re building @Hedginxyz while competing in the Colosseum Crypto World’s Fair Hackathon.
The goal is simple: make earning from Solana liquidity (especially on Meteora) more accessible. Deposit once, and a professional LP manages the positions for you — including hedging. Your capital stays in a smart contract with full transparency.
it integrate with @JupiterExchange and @PhoenixTrade too
The product is currently in beta testing. If you’re interested in hands-off yield or want to see what we’re building during the hackathon, follow @Hedginxyz and try the vault early.
→ hedgin.xyz, our x @hedginxyz
@MeteoraIDN @met_lparmy @MeteoraEco @MeteoraAG @vesper792
I don't think AI itself is the problem here. The real issue is the media trying to make everyone think it's dangerous.
xitter.cf/unusual_whales/status/…
Hot take: AI is in a bubble, and that doesn't mean it isn't the most important tech in human history, it is. The issue is it's winner-take-all. Cars created whole ecosystems around them, AI just absorbs the entire stack into one model.
A single Claude update can wipe out a pile of startups overnight. Meanwhile trillions are getting spread across thousands of AI companies getting billion-dollar valuations just for the word "AI" in their pitch deck. Most of that money is going to zero.
If that's your logic, just take an Uber instead of buying a car.
xitter.cf/gregpr07/status/210899…
I think the AI divide is genuinely getting bigger every day, and it's shaping up into a whole new class system. If you're not paying attention, you're already falling behind. Putting real time into learning AI right now feels like the only way to stay in the game.
xitter.cf/karpathy/status/210936…
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.
- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.
So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.


















