834 panels. 1 machine. 2 people on the ground. 1 standard 10-hour shift.
I can barely contain my excitement over what our AI model just pulled off on a live customer site. Even with full breaks, lunch, and morning stretch-and-flex built right into the 10-hour day, the setup didn't miss a beat.
We think this might actually be a world record for a robot+crew in a single 10hr shift.
@GrittRobotics
Seriously impressed with Gemini Neon - it's the first really usable model for coding by Google. It has clearer thinking and better judgement, especially around planning. I trust its thinking more than Fable 5.2 and Astra 6.2, but the harness is much worse than the competition.
1
54
Matt Jaynes retweeted
Actually, your granddaughter already has an account, even if her parents haven’t claimed it.
Nearly every (99%+) American child does. Over 90% are still unclaimed, but they will be claimed over time. Some in 2 minutes. Others in 2 days, 2 weeks, 2 years, or even 20 years.
And if a parent never claims it? Nothing is lost. No expiration. At 18, the account becomes hers. She verifies her identity and takes control.
The money is there in 25 million accounts, invested and compounding. It waits for the child. 🇺🇸
Replying to @MichaelDell
I wish there was some way I could get my granddaughter on this program. My daughter in law won't do anything that Trump helped promote. Serious TDS there. 😭 So my granddaughter has to miss out on this opportunity you folks are giving these kids. Thank you. These kids deserve so much more for their futures.
Same in systems engineering. Best systems I've seen are run by those that rarely even think about them. Their systems are simple and boring - totally unremarkable (by design). Sadly the worst engineers are often the loudest and promote complexity like it's a badge of honor.
Dating discourse is dominated by people who do a lot of dating which definitionally means they're failing at dating.
47
Scarcity can only be allocated three ways: price, queue, or lottery.
Pick your poison.
Datacenters are a globally competitive market. Someone, somewhere, will build them.
Replying to @SemiAnalysis_
A 100MW datacenter in China now costs under $300M to build, excluding IT equipment, about a fifth of the US cost. China's build cost has fallen about 25% since 2021. (4/6)
Matt Jaynes retweeted
Ah yes, communism the enlightened system that can only succeed if executed with absolute perfection by a caste of flawless, selfless, omniscient demigods who have somehow transcended every human flaw. Zero greed. Zero power hunger. Zero bureaucratic incompetence. Zero ability...
Matt Jaynes retweeted
Imagine delaying a cancer cure by 5 years because some people decided technological progress was moving too fast.
What we really need is more AI scientists. More automated laboratories. More medical research. More discoveries. More lives saved.
Accelerate.
Matt Jaynes retweeted
What's crazy is that music like this could have been played for centuries, just no-one thought to do it. What else are we missing?
The only repeated redirecting prompt I typically type or dictate from here is "could this be simpler?". Yields improvements every time. Seems like it should be part of the model system prompt by now.
Matt Jaynes retweeted
I used to be a language lawyer. In the late 80s I _studied_ the ARM (Annotated Reference Manual for C++). I knew the language cold, inside and out. I could make it do backflips. I shipped a lot of code using it. A LOT.
I’ve attacked other languages with a similar intensity and depth. Until recently, that was simply the only way to become adept at building software systems.
I now consider the deep understanding of a language to be a supreme waste of time and effort. I don’t want to learn Rust, or Golang, or Typescript, or any of the thousands of other languages out there. I’ll leave that to my agents.
What I need now is to understand the high level gestalt of the system. I want to see the systems and subsystems. I don’t need to inspect the cogs. If I truly need to understand a detail at the cog level, I’ll interrogate my agents. I’m not going to pretend to be a language expert anymore.
💯 The future is pruned, vendored dependencies with your agent doing periodic checks for vuln patches on the upstream and folding them in only when they apply to you.
Supply chain poisoning risk is too high to yolo your dependencies.
What’s going to happen to open source?
A prediction:
Dependencies are over.
The same Brotli vulnerability impacted .NET, Node, Bun, Envoy and more. Dependencies are too big of a target.
When rewriting a dependency is one prompt away, why take the risk?
86
Matt Jaynes retweeted
.@Jason is 1000% correct here. AI compute demand is going straight-line up.
We are just getting started.
Is this not obvious? It will be in hindsight.
I've been screaming about this for months.
Some models I've run.
(Wherever you think this is going, it not the green line. That's the last era of compute demand.)
“AI token usage is going 500x from here” says @Jason
Anyone arguing that AI usage has peaked or compute demand is maxed out is completely lost and obviously doesn’t use the technology
We’re still at something like 2% of people in corporate America using AI today AND both OpenAI and Anthropic haven’t even launched agents at scale yet
We have only touched the surface of what’s possible with AI and the usage will not just grow (linear) it will ACCELERATE (exponential) from here
I feel like this is hard for people to wrap their heads around but if you want to understand where this market is going, you need to understand this trend.
It’s not stopping, it’s getting much bigger, and the mid-terms, rates or anything else will not stop this
It doenst mean the market goes up in a straight line, but as an investor don’t get lost in the noise. This is the biggest technological change in our lifetime
Think MUCH MUCH BIGGER
Matt Jaynes retweeted
It's still such an adorable moment when the agent casually clicks the "I am not a robot" checkbox
You can't secure what you don't understand.
Yes, you can have your coding agents help you understand and verify things for you, but if you don't understand your systems enough to secure them, you are cooked.
Look at Railway, LiteLLM, OpenCode, Zeabur, TiDB, etc. All those guys are heavy users of coding agents and yet did not understand the basic vulnerabilities that got them p0wned. None of the exploits were from Zero-Day vulnerabilities - they were sloppy systems (mostly by allowing poor credential handling).
This is why **ruthless** simplicity is needed. Every hidden corner and dark place has the potential to get you p0wned. If you can't clearly see and reason about all your systems, you will be dead in this new environment.
Remember: No dark corners, no hidden areas, no yolo-hope-it-works when it comes to critical security.
You can't secure what you don't understand.
154






























