Eng @DRWTrading | Exited Founder | fmr @solstrategies @orangefincrypto (acq) @krakenfx | Opinions Are My Own
United States
Joined January 2012
- Tweets4.5K
- Following1.1K
- Followers6.7K
- Likes7.2K
Pinned Tweet
Bit of an announcement
Tomorrow marks my last day as CTO of @solstrategies
It's been an absolutely incredible ride. From starting @orangefincrypto , to coming onto SOL Strategies, to being the first Solana focused company ever to get listed on Nasdaq, to being the staking provider for an ETF, to so much more, it's been an experience I wouldn't trade for the world.
None of it would be possible without the amazing team I had the pleasure of interacting with. This includes @michael_hubbard , @Ehrls15 , @AfroMonkey404 , @coderigo , @TrilliumJohn and more that aren't on X, but appreciate them just as much. Thank you to all of you for being amazing to work with.
Additionally, I wanted to give a shout out to the @darklakefi team that have already brought such amazing energy and ideas to the company in @vitorpy , @DuchessofSOL , @yesjess , and the rest of the team. I'm super excited to see what the rest of you build!
Lastly, as far as the team goes, I'd like to also give a public shout out to @LeahWald , @TonyGuoga , and @moeadham . Thank you for bringing me on board, taking a chance on me, and teaching me many things I'll take with me for the rest of my life.
In addition to the team, I'd like to shout all the investors and ecosystem teams I have had the pleasure of working with. There's far too many to name, and because I don't want to risk forgetting to include one, I won't name any. But please know if I have ever talked to you, I appreciate you 🙏.
As for what's next, I'll continue to be involved in SOL Strategies as a consultant, helping to ensure everything operates smoothly and providing help in any place I can. Additionally, I plan on trying to find a happy balance between touching grass and experimenting with AI more 😀.
My thoughts on Solana and crypto overall remain unchanged. If you are a builder on Solana and need (free) help with anything such as intros, fundraising, technical advice, etc, don't hesitate to reach out. If there is a way I can help you, I intend to do my best to do so.
Thank you 🫶
Why does crypto not have its own QQQ / SPY with the same level of demand / PMF?
Is it the demand that's not there? Is it the products that aren't there? Is it that this is actually BTC? Or is it all, some, or none of the above?
Yes. All the alpha is in MCPs now.
Code is worth near $0. Data is becoming worth more valuable every day.
Don't waste your time with harnesses. Spend your time with data.
Max Kaplan retweeted
Underdog users will now be able to fund accounts with stablecoins, using AUTH 🤝
News from @nextpredictio: @Underdog today announced that users will be able to fund their accounts with stablecoins and crypto, in a new partnership with zerohash. ↓
Max Kaplan retweeted
Today we’re introducing AUTH, The trusted layer for onchain account linking
Simple for users, with the controls partners need.
Giving this 🧵 thing a try..
Today, I used my local LiteLLM to vibe code a new PHP app, utilizing GitLab CI/CD to deploy to railway with automatic cert renewal via eMudhra 🚀
🚨 CRITICAL: GitLab flaw (CVSS 10.0) affects 12 years of releases.
CVE-2026-110264 could allow unauthenticated remote code execution on self-managed servers.
Read - thehackernews.com/2026/10/cr…
I've worked remotely for the last 9 years. For the most part, every day from my home.
I have now worked in an office every day for the last 1.5 months, and I don't think I'll ever go back to working from my house again.
Max Kaplan retweeted
"The customer is always right". Solana's customer is traders. Private mempool usage ignores this golden rule
Yo @solana @PhoenixTrade you got that dawg in you?
I have seen 'You Can See Everything' twice now in theatres.
Second favorite movie of all time.
The 🐐of documentaries.
I'm not even sure though you can classify it as a documentary though.
It's a documentary, a comedy, and a drama all on one.
When I took 4 months off from working, this was actually one of the most fun projects I worked on.
I sequenced my entire genome so I could feed it into a private LLM. My intentions were really simple:
1. Get the raw data back
2. Feed it into a private LLM
3. Ask super high level questions such as "what are the top 3 best and worst traits of my genome?"
4. Go down every rabbit hole #3 led me down (this ended up being about 2 days worth of rabbit holes 🙃)
Many of the insights are personal that I don't feel like sharing, although overall I was pretty pleased about my results.
However, one I am willing to share was really interesting:
The past few times I have gotten my blood drawn, my bilirubin levels were elevated. To be honest, while I am pretty healthy, I had bigger fish to fry than worrying about this and my doctor told me not to worry about that. He said it could be the case I have something called "gilberts syndrome", which is a completely harmless genetic liver condition where the medical recommendation is literally to do nothing, so it was not even worth getting tested for it. I thought nothing of it for awhile because it was harmless.
Fast forward to when I started analyzing my genome, when I started asking the LLM questions about the worst parts of my genome I mentioned in #3, it actually flagged genetic variants consistent with Gilbert syndrome!
Again, my prompt for this was extremely high level... "What are the worst traits of my genome?" I never mentioned anything about bilirubin or gilberts syndrome because quite honestly, I wasn't even thinking about it. But going through this process strongly suggested that the root cause of the elevated bilirubin levels was Gilbert syndrome.
There were some other interesting things I learned about myself as mentioned, including diseases I am a carrier for but I don't have (helpful if you want to have kids), and overall it was a really cool experiment.
Overall, between sequencing my genome and token spend (Venice ai), it was about $350 all in, and a great learning experience 😁.
I can't find a way to use my genome.
I’m the most intense longevity athlete in the world and haven’t used my genome for a single thing.
Fully sequenced in 2012. Again this year, paying $7,902 for matched 30x long read + 100x short read.
Still, nothing.
What am I missing?
I've yet to try muse, but already bullish.
This guy is a DAWG. Love him
Alexandr Wang (@alexandr_wang) says he has more than 200 direct reports at Meta's superintelligence lab, and that it is by design:
"Yeah, I think more than 200."
"One of the design principles around this was making it very flat. Effectively, all the research that we hired into this group called TBD, they report directly to me."
"TBD is like literally anti-bureaucracy."
"So I don't need to be micromanaging you to do brilliant work. You're gonna be able to do that on your own."
"We have this kind of like pod structure where we have various pods and there's various technical leads of the pods who tie break on the technical decisions."
"From a quote unquote people management perspective or from a bureaucracy perspective, we're just anti-bureaucracy."
"I would be lying if I said I was the best manager to more than 200 people."
"They don't need managers. They need a great environment."
"They just need the room to cook."
- Alexandr Wang, Meta Chief AI Officer, in an interview with David Senra published October 11.
Max Kaplan retweeted
Got the keys to the first ever @unruggable_io office last month!
The last 2.5 years have been a mix of working remotely, building in @bill_papas_12's basement and spending time at the @Colosseum offices in the US.
Now we finally have a place of our own to build from every day!
Robotic arms, 3d printers, desks and full assembly station.
Here's a concept design of what we're working towards!
IMO the era of Hermes & OpenClaw is now over.
Over the last few months, ChatGPT has improved significantly and has nearly completely closed the gap on functionality of these tools.
The main reasons I used Hermes in the past was:
- MCP used to be horrible. I've used MCP servers that exposed hundreds of tools that would eat up so much context and dramatically degrade performance. CLIs are superior here, however it is time consuming and annoying to build a CLI for every service you need. Models and the protocol have improved dramatically and MCP isn't that much of an issue anymore. I would go so far to say it actually works great now, and CLIs are no longer worth the time investment.
- Codex is great and I used it every day, however until "work mode" was introduced into ChatGPT, you couldn't really use codex on mobile, whereas with hermes, you could. Now that ChatGPT has work mode, this problem is eliminated.
- For my personal use case, RAG is a huge quality of life improvement. If you have ever bought a piece of real estate, you know how much paper work is involved. Multiply that by multiple properties and multiple repairs and it gets really unwieldy to manage without something like RAG. Previously, I used qdrant hosted in my k8s cluster where hermes also lived. However, with OpenAI vector store & work mode in chatgpt (and a tiny self made plugin to connect the 2), I can do everything inside openai's ecosystem for a fraction of the cost of EKS.
The obvious con here is vendor lock in / data retention with OpenAI, but for personal use, I think this is fine.
By the time I am done this migration, I will reduce my costs by 10x ($200->$20/mo) and it's well worth it.
Seeing lots of posts like this lately. They are wrong. I use hermes every single day.
I own multiple rental properties on the side. The entire back office of my operation is managed and automated by hermes. This is a real, cash flowing business with customers and bills to pay.
While I don't use openclaw anymore, as I switched to hermes long ago, where I think a lot of people go wrong is thinking that hermes is a competitor to Codex / Claude code. It is not.
At a high level:
- Codex: I use to write all my code. 0 lines of code are written by my hermes agent.
- Hermes: Contains a bunch of CLIs built to operate my business. These include clis to talk to banks, process journal entries, automate the flow of cash, as well as move money in and out of multisigs (proposer privileges only) and brokerages.
Hermes is superior to codex in terms of UX when it comes to making a bunch of tool calls, crunching a bunch of data, and taking an action on it. Codex is superior for engineering.
In the list of things I am NOT going to do, I will not:
- Download some "Claude desktop app" or the chatgpt equivalent. This is because - 1. I use linux, none have a Linux app. 2. I'm not keeping my laptop open 24/7 (lol). 3. Its a complete joke to call hermes insecure but your running business on a laptop is secure (skill issue).
- Run on a "Mac mini" or some equivalent device. My whole stack sits in a very locked down environment with no public subnets, mounts a read only root file system so it can't just do "anything", real logging and alerting, and more. Mac minis are great and all to mess with, but 1. I don't use macOS at all and 2. I'm not operating a business of a f'n Mac mini.
- Use hermes as "an assistant." This was never the use case. Its way more powerful than that if you know what you are doing. You don't give an employee access to your email, so you shouldn't give your agent access to your email. Very common sense stuff.
I started running openclaw in the peak of the hype and switched to hermes shortly after. I'd say I actually use hermes more now than I did back then.
Most importantly, its made this business more profitable as I have been able to cancel multiple subscriptions, do not need to pay a bookkeeper, and more.