
Building something new. Ex-@trustmachinesco, ex-@GraniteBTC. The future is Bitcoin. Never Sell.
Joined October 2019
- Tweets829
- Following1.4K
- Followers1.3K
- Likes2.1K
Pinned Tweet
Thrilled that we can finally unveil @GraniteBTC after months of hard work!
I couldn't be more proud of the team and what we've built. We're excited to share our mission of unlocking Bitcoin's liquidity for the world!
Introducing Zomma Agent Fleets 🙂↕👇
It's 2026. Why is your customer ticket still stuck for days?
Because most teams still work in a single queue.
With Zomma, you can now dispatch a Fleet of parallel computer-use agents as soon as work comes in.
The biggest hack? Fleet = better outcome with fewer tokens. Each subagent gets just enough context to do the work well, saving tokens and time.
And you can score every subagent against your own eval set.
@ZommaLabs is in early access. If you're building agents for your operations, let's chat 📥
Blaize retweeted
Henry Ford’s factory workers earned $5/day salary.
At the time that salary bought 5oz of gold per month.
Henry Ford paid his factory workers~$20k/month salary in today’s dollars.
We don’t need a minimum wage, we need hard money.
Blaize retweeted
I've come up with a novel idea for how humanity should deal with AI risks.
It's powerful and deceptively simple. In fact, it can be stated in one sentence.
And it relates to something I learned from Elle Woods. (thread 🧵)
Blaize retweeted
We gave ChatGPT, Claude, and Grok control of a real Toyota Corolla 🚗, steering/gas/brakes, no human driving (just a foot over the brake).
Only one model was able to complete our entire driving course. Introducing DrivingBench. 🔥⌛️🏁
Blaize retweeted
Financial institutions paid $3.8B in AML fines last year while spending $206B on financial crime compliance, with labor its largest cost.
Today we're launching Zomma (@ycombinator S26)
AI operations employees for finance back offices.
We start with compliance: for every cross-border transaction alert, our agent works the case end to end on the same screens your analysts use: gathers the evidence, writes the assessment, drafts the RFI. No APIs. Onboarding is an SOP and a login.
92% on OSWorld. Live compliance-ops pilots with a US RIA and a $10B AUM wealthtech.
If you run compliance or financial-crime ops, or know someone who does, let's talk → zommalabs.com.
Blaize retweeted
If you had trusted the settled science of the day, here is how it would have gone.
1890s: you dose the baby with Bayer's new cough syrup, the non-addictive alternative to morphine, marketed for children. It was heroin. They named it after the German for heroic.
1900s: you rub calomel teething powder on your infant's gums. It is a mercury compound, and it leaves a generation with pink disease. Swollen, peeling hands and feet, screaming for months, and some of them died.
1910s: you cook in Crisco, launched by a soap and candle company, because cottonseed oil is modern and lard is what your mother used. The process that made it solid produced trans fat, banned outright a century later.
1920s: you drink Radithor, certified radioactive water, on prescription. Eben Byers took fourteen hundred bottles for his vitality. His jaw was removed in pieces and he was buried in a lead coffin.
1930s: a chemist dissolves a new sulfa drug in diethylene glycol, which is antifreeze. It kills 107 people, mostly children, and only then does America give the FDA power to demand safety testing.
1940s: your doctor lights a Camel in the surgery. More doctors smoke them than any other brand, and the advert runs in the Journal of the American Medical Association, which sold him the page.
1950s: DDT is sprayed over your kitchen, your garden and your children at school. The man who found its insecticidal properties has already been given a Nobel Prize.
1960s: your wife takes thalidomide for morning sickness because it is remarkably safe. More than ten thousand children are born with missing and shortened limbs.
1970s: the country puts down the butter and picks up the margarine, on the instruction of the American Heart Association. The fat everybody switched to was banned in 2015 as unsafe at any level.
1980s: the fat comes out of everything and sugar goes back in to make it edible. Obesity begins a climb it has never come off.
1990s: your father is prescribed OxyContin, because fewer than one per cent get addicted. The company later pleads guilty to criminal misbranding, and the overdose count runs into hundreds of thousands.
2000s: you take Vioxx for your knee. It is pulled in 2004, and the FDA's own safety officer estimates tens of thousands of excess heart attacks.
Not one of those was fringe. Every one had a professional body, a literature and a man in a white coat standing behind it.
Nobody who followed that advice was stupid. They were obedient, to the most qualified people available.
Settled is a word about money. It means the questions stopped being funded, so they stopped being asked.
So which of today's instructions will your grandchildren read out in disbelief.
The oil washed in hexane and built into every cell you own. The statin that blocks the pathway making your hormones, to prevent one heart attack per hundred people. The injection where a third of the loss is muscle. The infant formula built out of vegetable oil.
All settled. None of it funded to be otherwise.
Blaize retweeted
FINANCIAL GRAVITY:
If we divide the S&P 500 by the fed’s balance sheet, the line is basically flat since 2008.
The correlation coefficient between central bank quantitative easing and the price of stock indexes is nearly 1.
The money printed by the Fed, because of the structure of the Open Market Operations, is plugged directly into the Treasury markets, and from there, flows into equities and derivatives.
This has served to primarily enrich the asset owners, financial institutions, and wealthy elites who own the majority of the stock market anyways.
The entire rally has been an illusion, financed by the Fed and maintained through QE.
Wow, Fable is really expensive. 😅 Great output, but definitely need to limit it *only* to planning - even using it as a build coordinator blows through token budgets way too quickly.
And put strict guardrails for delegation: a surprise *team* of 29 Opus subagents chews through usage pretty quickly. 🤦♂️
Although the resulting product was pretty cool... :)
A reminder to everyone who is Fable binge-coding this weekend: use Fable as the brains but not the coding muscle, otherwise you'll blow through your plan. And it doesn't reset until Monday!
Promp:
A change to how we are developing going forward: you (Fable) are the planner / coordinator / orchestrator / judgement; cheaper models (Sonnet) are builder / coder / executor; use Opus as necessary.
Blaize retweeted
Everyone is still talking about Fable.
The bigger problem is that we've become dependent on cloud AI for everything.
That can't last forever.
Meet Friday: a personal AI assistant that runs locally and still works when you're 30,000 feet in the air with no Wi-Fi.
Introducing GLM-5.2: Frontier Intelligence, Open Weights
- Significant improvements in coding and agentic tasks
- Strong long-horizon capabilities with a 1M context window
- Two levels of reasoning effort: GLM-5.2 (max) pushes the limits, while GLM-5.2 (high) strikes a strong balance between performance and token efficiency
- MIT-licensed open weights
- Same API pricing as GLM-5.1
Tech Blog: z.ai/blog/glm-5.2
Weights: huggingface.co/zai-org/GLM-5…
API: docs.z.ai/guides/llm/glm-5.2
Coding Plan: z.ai/subscribe
Chat: chat.z.ai
Blaize retweeted
Introducing /visual-plan - a skill to generate rich, visual plans for Claude Code and Codex.
Plan mode in Claude Code is incredible.
But I always find my eyes glazing over when it gives me this huge markdown essay in my terminal.
I found I can make much better visual plans with reusable components.
So I made a skill called `/visual-plan`.
It generates plans as MDX with visual, interactive components. Diagrams, interactive API specs, schema design changes, annotated code, and even pan and zoomable wireframes.
So for any UI work, you can look at a wireframe first, comment on it, iterate, and then have the agent work.
I’ve found this to be a much more intuitive interface for reasoning about what the agent is doing.
It’s somewhat inspired by that popular post about how HTML is better than Markdown.
But HTML can be slow and verbose to write. And it doesn’t look good checked into a repo.
This has really made me feel like humans and engineering are entering a new abstraction phase, where we reason about things at the plan level.
As long as the plan is good, agents are getting more and more reliable at executing on it.
Almost to the degree that we trust the C compiler to compile to assembly reliably.
Plans are the new intermediate representation.
I also made a skill for the reverse of this, called `/visual-recap`.
After the agent works, it gives you a recap of everything it did.
Same idea: wireframes, interactive API specs and diffs, schemas, annotated code, etc.
So now when you’re reviewing what the agent did for you, or looking at a pull request of somebody else’s code, you can see a visual recap instead of just reading a wall of text.
It’s all free and open source. You can find it on my GitHub.
Will link to it in the reply because we all know how dumb these algorithms are with links.




























