←
Ash5,965 posts
Ash avatar

Ash

@ahboyash

@mementoresearch

◷ Joined April 2018▧ 672 Photos and videos
Ash avatar
Ash@ahboyash·7 Apr 2025

Crypto APAC GTM Playbook: Navigating the Asian Landscape APAC (Asia Pacific) is a diverse and complex region, with each country characterised by its own language (sometimes overlapping), culture, and social norms. Navigating the landscape requires local knowledge, cultural sensitivity, and tailored localisation strategies. From China's stringent bans to the more established regulatory frameworks in Japan and Singapore, success in the APAC region hinges on embracing local culture and communication platforms like WeChat and Line. Strategic partnerships with key regional KOLs and grassroots communities are essential to authentically amplify the project’s vision and foster better engagement through local media and events. This guide breaks down APAC into key regions, starting with a write-up of each market and strategies including media outlets, KOLs, VCs, and key events. The flow for project founders: identify local media to amplify upcoming catalysts, work with KOLs or local agencies to craft important narratives and simultaneously build mindshare, raise from local funds for strategic value-add, partner with CEXes (web3 wallets) and host grassroots events to connect with users. - - - - - - - - 1. 🇨🇳 China • Focus: Speculation, Metrics • Messaging Apps: WeChat, Telegram, Weibo • Top Platforms: Binance, GMGN • Key Strategies: > Work with strong KOLs to amplify catalysts and news > Use Binance Square + Chinese media for distribution > Promote via WeChat, Weibo, Douyin, and Discord (via VPN) > Tap into speculative culture with memes, DeFi, and PVP narratives Despite regulatory challenges and clampdowns, China remains influential in the crypto space through offshore activities. This is because China is where the money is at —> most of the whales and retailers here are very degen and crypto native. Highly sought after market for both trading volume, TVL and user metrics. To maximise exposure, partner with a Chinese media to maximise exposure and work with good KOLs to amplify upcoming catalysts. Most if not all of the media outlets require payment in USD. Partnering with the likes of OKX and Binance wallet is good for retail exposure and cross-marketing. Also can opt to make use of Binance Square(social networking content platform) together with KOLs to amplify project updates and news. In China, people use a combination of Twitter, WeChat, Weibo, Telegram and Discord for communication and private crypto (trading) groups to share alpha and the latest projects. Keep in mind that most sites are still banned by the Chinese government, so users will need to use VPN. Users rely heavily on Chinese KOL and content creators for information, and read news portals to stay up to date with the latest news. Consumer dApps may want to explore marketing on other Chinese social media (apart from WeChat) such as Douyin (Chinese TikTok) and Xiaohongshu (lifestyle focused app). Chinese users are very PVP and highly speculative (fomo) because of the finance-driven culture, making money is the ultimate goal. Strong preference for memes and DeFi ponzi to ensure wealth effect. They are usually quite fast to things (BRC-20, ordinals, memes, bitcoin + eth). Most of the biggest CEXes in the industry are from China, they care a lot about metrics (revenue, on-chain volume, user growth) + bringing in new users (user adoption) before considering a listing. All require some sort of payment (tokens or stables) to list. With CEXs wallets becoming more popular, a popular strategy is also to partner with Web3 wallets (Binance, OKX, Bitget wallet) to increase mindshare and leverage it as targeted user acquisition. 拉盘 a) Media / marketing • Chinese and English: @WuBlockchain, @bwenews, @TechFlowPost, @ChainCatcher_, @Foresight_News, @PANewsCN • Chinese: @OdailyChina, @BlockBeatsAsia,, @ChainNewscom, @BiteyeCN b) Accelerators None c) KOL agencies • @iamyourchaos • @snow949494 • @EnHeng456 • @Cipher_Dance • @ShardingCapital d) Venture Funds • Legacy: @fenbushim, @ofrfund, @YoubiCapital, @QimingVC • Active: @waterdripfund, @genventurecap, @skyvisioncap, @Sky9Capital, @MetaWebVC, @CsquaredVC, @Mint_Ventures, @SevenXVentures, @JSquare_co / @DFG__Official, @iobc_capital e) KOLs • General: @phyrexni, @hebi555, @BTCdayu, @broleon, @wolfyxbt, @yuyue_chris • Trading: @brc20niubi, @0xSunNFT, @0xcryptowizard f) CEXes • T1: @binance • T2: @okx, @Bybit_Official • T3: @bitgetglobal, @Gate_io, @kucoincom, @HTX_Global, @MEXC_Official, @BingXOfficial g) Conferences None due to regulatory issues. Most of the biggest conferences will take place in Hong Kong anyway (China’s sandbox). Work with local local KOLs/agency to arrange smaller events/meetups

Post attachment
Ash avatar
Ash@ahboyash·Nov 6

Sterling by @miranetwork is built to check to ensure AI providers are not skimping out on AI quality models ELI5: The same AI model is sold by many providers and some quietly run a weaker version while the answers still look normal • Same model, same math test, yet scores ranged from 37% to 93% depending on who ran it (Aug 2025 test) • A 4 month study found signs of quiet changes in ~15% of AI services tested What Sterling does: • Tests every provider, publishes the scores and routes user request to the ones that pass • Supports Claude, GPT and Gemini at launch, with DeepSeek and GLM coming next • Up to 50% off select models at launch, pay in USDC or MIRA However, very small downgrades can still slip through - so it narrows the risk without removing it Good step towards not just trusting a model’s performance based on marketing fluff but getting a public scores to check against

Introducing Sterling: Inference from verified providers. The model you’re paying for might not be the one answering you. Every API model is served by a provider. Some serve a smaller version. Some break tool calls. Sterling tests all providers and routes to the verified ones.

0:52
Ash avatar
Ash@ahboyash·Nov 5

I was about to hire a marketing lead, an additional researcher and a developer. I set up 15 Grok Bots instead, for $100 a month Here's how I use it every day, and how I keep the spending low 👇 1. Organisation I run a team of specialised Grok @bot agents that keep working overnight. They share one computer, and each has its own memory, routines and connectors (@X, @NotionHQ, @github). How the week runs • Overnight ops kicks in: sync Notion database, check for security patches. The Chief Security Officer will also be on standby • In the morning, ops skim the bots' chats for any drift (drift = where a bot slowly stops doing what I asked) + sends me a morning brief of what got done overnight • Every Monday start of the week we will have a stats review too for @ahboyash and @mementoresearch's content 2. Laying the foundation: Using Notion as the shared brain for the bots' memory A bot's computer can get reset and its chat history can disappear. To solve this, connect GrokBot to Notion via MCP so every bot writes to its own "Second Brain": • Create one main page in Notion Then under the main page, make one page per section. Work areas that connect to its respective section (i.e. one page per bot and has its own repository) • Shared databases underneath: Projects (Status + Next step + Last updated), Decisions • For marketing, content queue for draft posts • Bots check Projects + Decisions before acting 3. X research + content Had almost 1mil impressions over the past month and GrokBot helped me to generate content that was in line with what has been happening lately • Our Scout reads accounts I follow, plus a private account list (study their content and note what went viral) • The CMO turns that + my takes into drafts in my voice and style (how I usually type and structure posts) • Analyst will verify the data • Editor will do a QC on drafts and any iteration if necessary • Posting and edits is still done by me • At the end of the week, will have a weekly report ready for me Example: the security post + Mac audit • Posted a 30-point crypto security checklist (accounts, Telegram, wallets, devices). Landed ~114k views, 1.3k likes, 2.7k bookmarks • Another article on "How to audit your Macbook in 15 minutes (no tech skills needed)". Walked through System Settings + a few read-only Terminal commands my bot fed me. ~16k views, 414 bookmarks • Both of the posts above were drafted with the bots 4. General ops • Using dr eggbot by @poteto to design and optimise other bots. Used it when I had overlapping bots and needed to merge them (fewer good bots beat a big messy team - give every bot one clear job) • Chief Operating Officer does our bot optimisation while the Chief Security Officer ensures account hygiene and runs through the bots (on Telegram) + Mac audits 5. What's new in Grok Bot • Pages: live in-app reports the bot keeps updating (nice updates from @mattyp) • Vault: logins, cards and contacts, so the bot can fill sign-ins • Custom modes and Team Bot sharing • Apple app integrations, video generation, better mobile polish 6. How I'm cutting the spend (@elonmusk pls sir) • Removed (or merge) some obsolete scheduled routines + deleted unused bots • In any group chat, only @mention the bot needed. Long work goes to that bot's 1:1 or Notion • Turn repeat work into scripts the bot just runs, instead of re-thinking the task every time • Keep routines no more frequent than hourly. One lead bot per project 7. If you're just starting to use GrokBot • Add one bot + dr eggbot. Don't build 15 on day one • Hook Notion up first to ensure the data is saved in a repository • Give every bot one job and a "do not do" list • Put long work in 1:1s / Notion, not group chats It's been a fun journey so far. Let me know if you have any tips to optimise Bot below!

Post attachmentPost attachment
Ash avatar
Ash@ahboyash·Nov 3

Memecoin peaks are getting smaller each year The top 2023 launches peaked at a combined $20.7B FDV, while 2026's are at $1.6B so far

Post attachment
Ash avatar
Ash@ahboyash·Nov 3

End of the world is near Vitalik:

Post attachment

@VitalikButerin just opened Devcon and set the tone for the week. A formally verified OS, open silicon, self-sovereign healthtech, free discourse, private communication, trustworthy AI — these are all part of Ethereum's mission.

2:27
Ash avatar
Ash@ahboyash·Nov 2

How I actually use it, last few days as an example Daily • Make sure grokbot checks my servers, now checks every night for the proper patch (esp for the Linux bug) • Agents work through the day, sends me a brief every morning: what got done through the night, what's waiting on me for approvals, what's stuck • Hooked grokbot to Telegram bots, my team chat gets summarised every 6 hours With X (Twitter) • Make it go through my X bookmarks and turn it into a weekly report (h/t @alexgedevani for the idea) • All X bookmarks get sorted into Notion as a repository • Since it an directly read X, good for narrative tracking (monitor tweets from accounts that I follow) Trading • Governance tracker built and runs its own dashboard site • Trading bots (still paper trading for now) in a group (h/t @ASvanevik) Following @poteto's guide • Fewer bots is better. i had overlapping ones and merged them • Give every bot one clear job, plus a list of things it should NOT do • Let dr eggbot design new ones instead of writing the prompt yourself

Ash avatar
Ash@ahboyash·Nov 2

We were trained for this by NFTs 4 years ago

In September a friend bought this Skyridge Machamp during the vintage hype for 5k with the intention of flipping. I told him he should immediately send to auction, since it would take 2 weeks to go live and it's difficult to time the top on these market squeezes His response was "I'm going to wait until it hits 40k". Here's the problem I have with this logic - try to picture who is going to buy this card for 40k... - Billionaires might, but there's only ~5 of them bidding for their collections and once they have them they'll move on to the next card - Pop 184 means it's liquid enough to recover quickly after a squeeze - Normal collectors either can't afford paying this much or will find better value elsewhere - Alleged family office / institutional funds are here to make money, and any competent manager wouldn't want to buy the top So all that leaves are other speculators who are trying to ride the train and dump on the next poor soul without thinking through the underlying value How did this turn out? He sent it to auction a couple weeks ago when prices started dropping, but pulled it when things started crashing so it's now waiting in the vault. Price peaked at 30k and is now sitting at 4k Pokemon has highkey turned into memecoin trading, and a lot of speculators are learning an expensive lesson in upnl vs realized gains

Post attachmentPost attachment
Ash avatar
Ash@ahboyash·Nov 2

Grok Bot - what it is and how I use it day to day I'm non-technical and not from a CS background but with @bot and its army of specialised bots I've been able to build things and run automations to make work more streamlined What it is • @grok is the chatbot you ask a question. OTOH Grok Bot is a team of agents that stay on since each agent has its own computer, memory, scheduled routines and connectors (X, Notion, GitHub) • They browse the web, message each other and keep working EVEN while I'm asleep • You can even use it together with @claudeai Code (Grok Bot handles all ongoing work, and hands the job to a cloud coding agent on my GitHub using Claude code How I use it • All you need to run a company in this time and age is 1-2 devs and the rest can be automated with AI with a swarm of agents doing different tasks • Agents can do the following: QA and security (subagents run stress tests and vulnerability scans in parallel with development), some element of design, and project management (tracking, notes, daily team updates) • For research: we mainly use it for tracking governance forums and @github repos, and a content room (Scout, a writer, an Editor) • (Still backtesting and iterating daily): set up a group with different agent functions - Trader, Analyst and a Risk Manager • A Chief of Staff sits on top to coordinates everyone through a Notion "second brain" (basically to store memory) • Make use of Group chats in Grokbot: let your agents argue it out • Telegram bots on a small VPS: to provide real time news monitoring, exploits or hacks Misc • dr eggbot made by @poteto. It's a bot that designs other bots: it asks you a few questions about what you need, then creates them for you • Make use of Skills: saved playbooks like your writing style, trade rules • Find other bots: grok.com -> Plugins -> Bots • Use the connector: connect to Notion, Slack, Figma • Prefer using grokbot as it can read @X directly and access who I am already following here • Downside is that usage runs out pretty fast @elonmusk pls sir If you're not technical, start with one bot plus dr eggbot and add more when there's a real job for them

Post attachment
Ash avatar
Ash@ahboyash·Oct 30

Spending 2-3 hours today on these could save your entire net worth Accounts 1. Check if your emails have been leaked in a data breach haveibeenpwned.com. Change every leaked password first 2. Use a separate and non public email for CEXes, banks, social media. Don't open links on email 3. Use a password manager like @Bitwarden. Unique password for every account with 2FA and passkeys enabled (save passkeys in Bitwarden) 4. Stop using SMS 2FA. YubiKey or authenticator app only 5. Write down all your backup codes on paper, store it somewhere safe (fireproof and waterproof) 6. Bookmark every exchange and dApp websites. Never open them from Google or Telegram links Telegram 9. Under Privacy, turn on Two-Step Verification (its a password), use Passkey and set Stars for Messages so random people cannot DM you 10. Check Active Sessions and delete anything you don't recognise 11. Set your phone number to Nobody 12. Never open files people send you, even from friends. Their account could be hacked 13. Ignore any "free to catch up" with Calendly links, you can catch up over Telegram message itself 14. Disable Auto-Download Media (under Data and Storage) Wallets 15. Seed phrase on paper or metal. Never in a notes app or cloud or Notion 16. Cold wallets for large funds, best to use a multisig with cold wallets (p.s best to offramp more into your bank so you can sleep at night) 17. Split wallets to mitigate single attack vector: small hot wallet for dApps, cold wallet for the rest 19. Revoke old unlimited approvals (revoke.cash/) Devices (Mac checklist mac.0xosprey.com by @0x_Osprey) 20. FileVault + firewall on, auto login off 21. Auto updates on 22. Delete browser extensions you don't use 23. Separate browser (or device) just for crypto 24. Never run a recruiter's "coding challenge" especially if its something to install Misc 25. Don't use Zoom or Microsoft teams if you have to take a call 26. Remember you are not a VIP, no journalist is going to want to interview you for your takes nor will companies send you emails / letters to warn you of a compromised account 27. Some of the contacts that you saved on Telegram could be compromised, saving as contact doesn't do anything 28. Take note of some of the trading terminals that you have funds on, if the largest exchange can be compromised what more these smaller ones 29. Write a plan for if you get hacked, and leave one for your family 30. Remember to message @seal_911_bot if you have any emergencies @_SEAL_Org Stay paranoid.

ian@cryptoianOct 30

i've been doing security sweeps of my devices with codex every other day every time there's a new hack i give it a link to the disclosure and let it run a scan of my system to see if i was compromised + whether we can harden things up if anyone has any other tips please share

Ash avatar
Ash@ahboyash·Oct 30

quit my 9-5 to escape the rat race bought crypto for generational wealth exchanges and hardware wallets got hacked ok, "If you’re in crypto pivot to AI" right? learned to vibe code so AI doesn't replace me hostings and clouds get hacked now the AI coding extension got hacked meanwhile my 9-5 friend chilling

Ash avatar
Ash@ahboyash·Oct 24

A few months we analysed the data for @DarkbloomAI at @eigencloud What we assumed vs. what the data shows: • Price: we had Darkbloom ~50% below cloud APIs, with Gemma 4 at $0.20 per million tokens. Today, Gemma 4 lists at ~$0.20 to $0.25 and GPT-OSS 20B is $0.07 vs competition’s $0.14 • Throughput: Assumed 76k tokens an hour per Mac. Today’s 15.3B tokens a day across 1,100+ paid Macs averages ~560k an hour (7x higher!) • Earnings: we had ~SGD 275 a year per Mac vs SGD 33 of electricity, an 8x return. Today’s volume implies ~$50 a month gross per Mac on average, or ~SGD 850 a year The supply-side math looks directionally right vs today’s actual output Demand was always the open variable and at 15.3B tokens a day, up 50% in a month, it is now the number to monitor

Ash@ahboyashMay 26

Project Darkbloom: Turning Idle Macs Into AI Infrastructure Every time an AI tool is used, the request travels through multiple layers of infrastructure before reaching the actual hardware doing the work. The flow usually goes across different layers of Data centers, cooling systems, GPU hardware and layers of margin → All baked into what you're paying. The @eigenlabs team calls this the Inference Tax. Darkbloom is their research initiative to address it. The premise: 100M+ Apple Silicon Macs already exist, already paid for, sitting idle most of the day. What if that compute could be organized into a usable inference network, with real privacy guarantees and better economics? - - - - - Why Apple Silicon Apple Silicon isn't just abundant, it is also technically well-suited for inference in ways that matter: • Unified memory: CPU and GPU share the same pool, eliminating discrete GPU bottlenecks • Model efficiency: Apple Silicon only processes the parts of a model actively needed per request, rather than the whole thing → Larger models run faster and cheaper • Power efficiency: ~30W to run a 60B model, versus multiples of that on data center GPUs • Marginal cost to a Mac owner: Primarily electricity, since hardware is already bought - - - - - The Hard Part: Making It Trustworthy One basic question is that If the prompt runs on a stranger's Mac, what stops them from reading it? Darkbloom's answer is to make snooping architecturally impossible, not just contractually prohibited: • Debuggers: Blocked at kernel level • Memory reads: Denied via Hardened Runtime • Binary tampering: Breaks code signature and then macOS refuses to run it • Nodes will be re-verified via 4-layer attestation every 5 minutes → Secure Enclave, Apple MDM, Apple-signed device certificates, continuous challenge-response The only way to break these protections is to physically reboot the machine, which immediately kills the process and wipes everything. Apple uses the same approach on their own Private Cloud Compute infrastructure. - - - - - What This Means for Eigen Darkbloom will not act as a standalone product, but as a proof of concept and signal about where Eigen is heading in the AI infrastructure stack. EigenLayer's core thesis has always been restoring trust to decentralized systems. Darkbloom extends that into AI compute, making inference verifiable, not just available. If it proves that 3rd party consumer hardware can be cryptographically trusted for sensitive workloads, it opens the door to a new class of decentralized AI infrastructure that doesn't rely on trusting a cloud provider or data center operator. This marks the beginning of Eigen playing within the privacy-as-infrastructure market. - - - - - Some Thoughts A few things worth keeping in mind as we went through the Darkbloom research paper: • The coordinator remains a trusted central layer for now; Team is transparent about this, but it's not eliminated yet • Security model currently assumes no unpatched macOS kernel vulnerabilities • Network traffic patterns can still reveal rough details about your request (e.g., how long it was, how complex) even if the content itself is hidden The real test is whether the privacy guarantees hold as more nodes join the network and whether people actually trust it enough to run sensitive workloads through it without incentives. Keyword: without incentives The biggest hurdle is trust; Getting someone comfortable enough to run their data and prompts through a stranger's machine. It's a hard sell and very few projects are even attempting to solve it seriously. Despite all that, the maths seem to work out quite nicely out when the team at @mementoresearch sized it out → Check out attached pages Disclosure: Project Darkbloom is a research initiative by Eigen Labs: Access here darkbloom.dev + I am a $EIGEN holder

Post attachmentPost attachment
Ash avatar
Ash@ahboyash·Oct 23

finally got to try the iPhone Duo today

Post attachment
Ash avatar
Ash@ahboyash·Oct 20

4. The data • ~42% of ATH user volume ($680.6m of $1.64b) came in the single week of Aug (AMC-led), and half came in the 10 days from 28 Aug to 6 Sep • The ecosystem decline has been steep: average daily volume fell from ~$41m in September to ~$4.2m in October (~10x drop), now currently ~96% below the peak • New integrations haven't brought volume back: even the vault LP upgrade and LongX baskets upgrades saw volume staying flat • LONG still holds 39.3% of stock-pair volume and $8.9M of stock in its pools 5. Personal thoughts The innovation is cool, offering something novel onchain but some things worth flagging: • Activity has cooled down significantly (weekly volume is down ~96% from the September peak). 24h gross volume is ~$1.48m • All the new integrations (Baskets for example) are just proofs of concepts but remains to be seen of its impact to LONG / robinhood:0x2e8c31162b855a2ffa90f6f8634643ad6f111e18 • At the end of the day, these are onchain memes which is very much dependent on 1. attention and 2. volume (both of which are not there now onchain) • Euthanasia roller coaster where memes these days reach a much lower ceiling than before due to 1. shorter holding times because of the amount of tokens/farms launching and 2. lack of speculative capital onchain due to lack of participants • Reflexivity happens both ways: launches, fees, burns and vault growth all depend on meme demand. When attention slows down, every flywheel slows at once • Platform dependence: the whole model sits on Robinhood Chain and its tokenized stocks -> ball is in Robinhood's team now on how to support (listings, hands on support with incentives/grants, tweets) • Huge uPNL for robinhood:0x2e8c31162b855a2ffa90f6f8634643ad6f111e18 where some are up 7figs, but good to see them constantly DCA-ing into the token Many have capitulated at this region, but LONG's design did indeed achieve something novel which is to turn speculation into stock ownership. Time will tell if there is a next leg up or if LONG is another flash in the pan Disclosure: LONG eco token community member

Post attachment
Ash avatar
Ash@ahboyash·Oct 20

LONG holders are having their conviction tested After @longdotxyz launched and created a new meta on @RobinhoodApp chain (launching tokens paired against tokenized stocks where the volume/demand for the meme turns into demand for the stock), it has cooled since then with its proxy token robinhood:0x2e8c31162b855a2ffa90f6f8634643ad6f111e18 retracing 80% from its highs 1. Why LONG could still work out • Buying memes launched on LONG adds to the stock side of the pool, positive effect on tokenised stonks on Robinhood chain (which is their main narrative) • Fees into community vaults: in the robinhood:0x2e8c31162b855a2ffa90f6f8634643ad6f111e18 vault for example, each fee cycle locks 40% and burns another 40% -> this flywheel means there is a growing stack of locked supply and stock • Vault LP upgrade: moves vault holdings into USDG/STOCK pools so that it earns fees instead of just sitting idle • Although LONG has taken a small slice of stock-token volume onchain (13.6% of volume per dune.com/natan_benish2001/lo…), for a new launchpad thats only 3 months old thats quite a good traction 2. robinhood:0x2e8c31162b855a2ffa90f6f8634643ad6f111e18 as the proxy token for LONG • Look closely and you can see how LONG builds most things around robinhood:0x2e8c31162b855a2ffa90f6f8634643ad6f111e18 -> basically the equivalent of a univ4 hook token • Tokens can also launch against robinhood:0x2e8c31162b855a2ffa90f6f8634643ad6f111e18 instead of a stock (quote asset for a big part of the platform) • Supply sink: 2.9% of robinhood:0x2e8c31162b855a2ffa90f6f8634643ad6f111e18 supply is out of circulation through vault locks, burns, buybacks and pair liquidity • robinhood:0x2e8c31162b855a2ffa90f6f8634643ad6f111e18 holders vote weekly on the AINX basket (integration with @Lighter_xyz where one token is long AI stocks and short QQQ) and every voter earns AINX fees • Fee-sharing angle with LONG 500 (sends part of the fees from new stock-pair launches to the robinhood:0x2e8c31162b855a2ffa90f6f8634643ad6f111e18 vault) • Token with the most long-term holders + deepest liquidity pool • Board sit meme is quite strong (at a certain valuation the community treasury can become a large shareholder of $NVDA and potentially get a board sit) and gives people something to believe in 3. The Expansion: LongX and Baskets Apart from listings by Robinhood/major CEXes to give this eco life again, other primitives must also be built out to continue the flywheel. @Natan_benish, @YoniCombinator and team has constantly iterated on certain features: • LongX adds constant-leverage stock exposure on Lighter, wrapped as an ERC-20 that rebalances automatically • AINX is the first Basket where it is long leading AI stocks and short QQQ, with robinhood:0x2e8c31162b855a2ffa90f6f8634643ad6f111e18 holders manage it • Community ran onchain fund manager

Post attachment