Kyle Mistele 🏴☠️
@0xblacklight
Clanker Wrangler / CTO @humanlayer_dev
My other computer is your computer
1.3K Following 7.5K Followers
But they were all of them deceived, for another watch was made
Sam Altman @sama just ordered 7 ultra-rare custom Swiss watches from Vanguart, fully branded with the OpenAI logo on the dial. One for him, six for his inner circle.
Inscribed on the back of each watch is ‘if AGI.aligned: deploy()’ which was written in the programming language Python.VANGUART
This is starting to be a trend in the tech world and I’m all for it
—
If your company wants customer branded Rolex or other watches shoot me a dm we specialize in bulk corporate orders with custom engraving and dial branding
Show more
there are literally “software engineers” out there rn who think you don’t have to read code anymore
the bar is literally on the floor
it's funny how I hated my ARM assembly class in college (to be fair the instructor and TA's were quite bad) but i look back on it very fondly
i really miss writing code by hand :(
writing quicksort by hand from scratch seemed awful at the time but being so close to the control flow and memory was amazing
Show more
the only way to know what's in the PR, is to merge the PR
- some of y'all
crazy that locale doesn't do some super lightweight inference to guess what I would like each week - could even do this with some basic clustering or multivariate analysis without LLMs
the recommendations are not great but we have a super consistent pattern of what we order the recommendations are never good
Show more
lots of coding agent memory systems are unhelpful - at best
i have said this a lot and some folks think I'm against agent memory in entirety
which is not true -
I just think that users should be on the loop that generates memories
and should be able to enable / disable the system or individual memories at will
and should be able to dismiss disruptive or unhelpful memories
putting a user on the loop also lets you build a better data flywheel over time to improve the quality of recommendations
live in @humanlayer_dev btw lmk if you want beta access
Show more
lots of coding agent memory systems are unhelpful - at best
i have said this a lot and some folks think I'm against agent memory in entirety
which is not true -
I just think that users should be on the loop that generates memories
and should be able to enable / disable the system or individual memories at will
and should be able to dismiss disruptive or unhelpful memories
putting a user on the loop also lets you build a better data flywheel over time to improve the quality of recommendations
live in @humanlayer_dev btw lmk if you want beta access
Show more
an S3 WAL-based filesystem fuse-mounted and configured as system swap space for downloading more RAM is anybody building this
so @jhleath doesn't want you to know this but you can just mount an archil disk to your machine and configure your system to use the mount point as additional swap space
allowing you to download more ram
on S3
Show more
the battle against slop is never-ending and oxlint is my best warrior
Nice this is really cool
New blog post! Write-Ahead Logs on S3: Compatibility Is Not Performance. We measured S3 Express against S3, then against Tigris and of course Cloudflare R2.
the superintelligent bash homunculus remains undefeated at chrome profiling & performance optimization
the only downside is my machine feels haunted with all the windows popping open and closed
Show more
WAL Street
lol i think dex has been mucking around in my agents config
Thanks for taking my rather too on-the-edge tweet and inviting a serious debate - I much prefer this.
I'm interested by how you define the term "coding".
I'll just @JohnOusterhout's definition of Tactical vs Strategic programming.
Tactical: the on-the-ground, day-to-day aspects of coding.
Strategic: the long-term. Codebase health, architectural design. Making the right decisions.
I agree that AI has largely eaten tactical programming. But I've seen no evidence that it can think strategically.
Ousterhout talks about some employees as "tactical tornadoes" - able to churn out astonishing amounts of work with zero eye for the future.
That's how agents feel to me right now.
I'd love to see how your observations fit into that frame.
Show more
I think there’s an important distinction to make here which is that models can do things FASTER than humans
But not always BETTER
Maybe this won’t always be true but I have never had a model come up with a good design for distributed systems problems or abstractions
It takes so much steering to get them to do good work in these areas
Maybe they seem smarter because they can come up with 4 options + pros/cons faster than humans but usually none of them are the most-correct answer
Show more
tbh this is a good debate to have.
Here is what the timeline of the near past/future looks like to me:
1. Models are able to code better than I can
2. Models are able to do coding-adjacent engineering work better than I can: debugging, profiling, optimizing, bug fixing, system design, abstraction design, ui design, idea generation, etc.
3. Models are able to do most things that can be done on a computer better than most people
For the kinds of coding that I do, Claude has achieved 1 and some (but not all) of 2. It is showing early-ish signs of 3.
This is not the case for all coding yet. If you are Anders Hejlsberg and are a world class expert in compilers and type systems, Claude might not be at 1 yet. My bet would be that it gets there quite soon, while at the same time pushing further into 2-3 territory for the rest of us.
You can think of model capabilities as capturing the distribution of human ability. I consider myself an average programmer so Claude has already surpassed me, but the same is not yet true for everyone.
Back in Nov 2025 I fully stopped writing code by hand, though I continue to code (using agents) every day. That is the point where for me and many others around me, it felt fair to say that coding is solved. But engineering is more than coding. Claude’s code is not perfect. It has bugs and inefficiencies. Claude can code, but it cannot yet do everything that goes into engineering.
Opus 4.8 is the first model that felt like it could code better than me. Fable now routinely finds optimizations and debugs issues that I would not have been able to fix myself. And with each generation, the code the model produces continues to improve, and the model is able to do more of the non-coding parts of engineering.
Show more
I wasn't planning on it originally but implementing Workflows, Queues, KV on celld is pretty easy now that Durable Object primitive is working well.
it would be a little silly not to do it now...
“Look at my incredible new factory!”
Yo that’s cool, what do you make?
“It’s highly optimised, fully automated, zero tolerance for defects and with a continuous feedback cycle”
Cool cool, so what do you actually make?
“I can interact with it on my phone, laptop, messenger, completely async, and the shared context means it’s always learning how to get better”
Very impressive, but what do you make?
“Every agent has full context, can spawn other agents, review their work, fix defects, and ship continuously.”
yes yes. WHAT DOES IT MAKE?
“Software.”
Oh nice. What software?
“Well right now we’re mostly using it to improve the factory.”
Improve it to make what?
“Anything!”
Such as?
“…a better factory.”
Show more
i measured the codemode implementation inside of executor vs opencode2 and here's what i found:
Working in HumanLayer
i guess we doin streams now
