kami23

Or it's just an experiment to share, I get what you're saying but not everything needs to be feature complete to share.

I actually vastly prefer people show me their broken demos. Let's me understand how difficult the problem is than I think initially.

We started using the ADK before the go sdk was available and went pretty far with using the python one. I hated it in so far as dep management is still a nightmare in python for me. I'm sure I was just holding it wrong, but it was getting to me.

I threw Claude at it to migrate to go sdk finally and had it AB testing it all in a few days. Stuff like that would have taken me forever in the past.

Now to attack some stuff that's in niche as hell jvm languages.

Our running joke last year was that anything we were learning or implementing in regards to AI was going to be outdated in two weeks.

Every week we'd build a new thing in our agent workflow and then Google's ADK the following week would have support for something we could have used instead of rolling our own.

Reading this release makes me wonder what makes it special, nothing sticks out from the press release.

But I agree, the churn is real, people are starting to throw around the term software factory at the management level and I fear big efforts there will be outdated by the time they finish, even with coding agents support.

Methotrexate is such a gnarly drug, my partner was on it for rheumatoid arthritis for a little bit, but the side effects were too awful for her to continue it. Luckily there was alternatives. Didn't know it was used in chemo!

I like this RCA a lot more than most. I appreciate the ownership and clear explanation of why their testing did not catch this bug.

Also impressed with the speed to address the issue unplug the model and then pretty quick remediation, the speed to remediation matters less to me, but good to see they can act quickly on their platform to kill the serving of the model.

I think in the last year I've had it happen once where I seemingly got someone else's response, it was very odd. I rewound and got what I expected. It was correlated to Anthropic's roll out of new models at the time which always seems to cause instability in their apis.

I've been toying around with something like this using jev for routing and having precanned components to display similar to a2ui but I think I'll take a new approach similar to yours. Thanks for sharing.

I bet they would be able to figure out a lot through game guides over the years, but not sure how popular those are. I play a lot of search action games and still look things up a lot.

I've been using the same technique. If I describe precisely the technical steps to accomplish the thing I need with no mention of the overall use it happily generates the solution.

I think I've had more success recently with telling it to look at an old project I have in another folder on the system that does the thing I need to do. I've been using that technique to try and build a little self replicating worm to see it work in its own world. I keep jokingly telling people that we should just let a group on the 'good' side, whatever that might be, exploit and patch and move on. Inspired by Linux.Wifatch https://en.wikipedia.org/wiki/Linux.Wifatch

At first I was like what a silly contraption, and now i just appreciate how straight-forward the solution is. I ended up wiring components onto relays to do power management on a past project and using a pi to control it, but in hindsight I was over thinking it. This would have solved my problem a hell of a lot more safely than what I did.

Whoever is doing these infiltration are still out there and nothing is really changing with how these are responded to, these guys aren't going to be the only ones. The cat is out of the bag, you gave people the access to unlimited automation, and it turns out people know how to use it for both 'good' and 'bad' for your various definitions of the two.

Defense as we see it today seems fundamentally broken. The answer is to fight it with more AI? That doesn't seem right. Can software reliably self evolve and keep patching itself? What certification is going to protect you from an intruder with their legion of minions, none of them, you don't have perfect visibility to your systems, your systems emit too many signals to ingest 100% of actions, so does software need to get fundamentally simpler?

Today has me wondering how many datacenters exist out there just for organized crime? Are the big kingpins of the world going all in AI as well?

Lots of questions popping into my head today, I'm trying to think how you would slow or stop this and I can't help but think we just need everyone to go outside and touch some grass and remember we're all stuck on the same rock.

You are going to revolutionize my life. I was trying to think of cool things to build for my tablet as well as now aiming for specific quirks of certain models is no longer an obstacle and their older custom android forks.

I'll be sending architecture diagrams to my agent, I draw those much faster than I talk. I hope there's a good way to get alerted for agent waiting for your turn. Hmm I wonder if the Claude Code app will install on my Boox. I've been finding it fun to run Claude on weird things, if it runs linux there's a good chance. My favorite install was into my Steamdeck to debug some button sensitivity issues.

Love that, I have similar desires to be able to control some climate controllers so I can get a more native bluetooth connection to it and override their programming for fine control of devices and have it never phone back home. One of these days I'll have time to throw an LLM at it, but I've been working on a dead project for 4 years ago I dropped because I realized how daunting it was going to be turn it into a product, but now I've made progress that would have taken me a year, in a few weekends.

They don't even let you really run those yourself now either, We moved off a big GHE footprint because the support for it was getting abysmal and features were slowly become GH.com. Ive been dreading the move to GitHub.com and it was worse than I expected.

Coder uses terraform providers to do the provisioning. I think that's the correct layer of abstraction, all major providers will be there and can be customized with the modifications needed via various other terraform features.

I've written three different solutions for platforms that solve this standard protocol layer. What has worked for us was deciding to standardize on the Kubernetes API for our services back in 2017, and then from then on, all our providers have provided either a k8s API to interact with resources, or provide a Cluster API for provisioning. For example we have two datacenter VM providers that implement the k8s API and we've been able to swap the providers with no user refactoring, it's a good solution in hindsight with how many people have k8s in their stack.

For the second big feature you're thinking of, we never really go a good solution to this problem, a lot of our workloads are long lived and not necessarily spot instance-able, but it's a very fun routing problem. Most of our need to be in multiple clouds is that we want to heavily separate our customer facing services, and keeping it away from internal workloads that support those customer APIs.

Is this public work? I'd love to take a look. This is my bread and butter.

I've noticed a few self identify and other random occupations, cool to see those fields checking out this tech at a deeper surface level.

I'm just a dev, but I appreciate the insights from other professionals.

You should have seen some of the Flash sites people made in the mid 90s early aughts. They all seemed to be straight up screenshots of the desired website and then buttons stapled on random portions.

They were an accessibility nightmare, but you use what you got. I tried so hard as a kid to understand flash, but had to settle on MS Frontpage to publish my first RPG page.

What's old is new again.

Ah I interpreted it as 'of course they can't stop distillation if they couldn't stop a model from escaping its sandbox'

I can see how there's a big leap there, but I agree somewhat. If they are aware these are happening and can detect it as it is happening why are they not stopping them? What do you do there? It'll be cat and mouse for a while. Thinking of reasons they wouldn't try and stop it is just a lot of speculation in my brain.

It's probably a way harder problem than I think it is, but they are aware of them now, so I assume they are going to get more aggressive about it.

Let's say then that they can't detect them near real time or even a bit after, maybe they do have a big observabilty gap that no one has solved adequately.

The speed which they add features I've needed for governance is pretty close to the speed I 'manually' write those for my company. To me personally we are all just going fast and breaking everything and not having enough time to set up safe environments. I'm sure it's in the backlog.

That they should be able to find distillation 'attacks' if they had enough observability.

Kratom has been around for a while, I remember seeing it in headshops at least a decade ago.

I remember watching GigaOm on Revision3 back in the day, I think I stumbled upon it via Diggnation... nope I think it was CrankyGeeks actually.

It was the first newsletter I actually subscribed to back in the day! Sad to hear about his passing, his appearances on podcasts introduced me to the more business side of tech where I was just a hobbyist teen at the time.

RIP

I haven't had time to read the whole article, but I really appreciate the cross section of the world that reads HackerNews and plays STS2. STS1 and STS2 are my favorite games and to see this pop up here brought a big smile on my face. Thanks for sharing.

I have this same attitude. I've been a linux user for 24 some years I don't need to know why Linux broke, just fix x and move on.

I used to spend hours debugging video card issues and other modifications I've liked to make over the years and being able to describe my ideal system admin setup I could get onto what I actually wanted to do.

Heh, thinking about it now, I broke a MBR on a Windows install as a kid and if I would have had these tools I would have been able to fix it immediately, but back then it took me using enough Linux booted off live cds to learn debugging techniques to fix the MBR. And debugging is one of my best skills.

Seeing subagents working in Claude last summer, I saw it and told myself my job is going to be different and I can automate the hell out of my workflow

This read like poetry to me. Thank you for sharing it.

I have a linguistics background and a lot of my philosophizing lately has been on whether or not the emergent abilities of the LLMs is deep down a similar mechanism that creates our consciousness.

For a little bit I was working on having linguistics based evals for a kaggle competition. My challenge was whether or not I could mask things well enough to not trigger its internal state of certain phenomena, and that sent me down a rabbit hole that I'm still exploring.

This story resonated with a lot of questions that can come out of figuring a good solid answer to the what is consciousness question. The one I triggered for me is: Is our perception of time just a slow thread in the giant GPU we are running the universe on? Or more generally, what is time? That's a fun YouTube rabbit hole if you ever need one.

I am unsure what you mean by hard evidence in the context of AI then, what is the evidence we are negating in your view?

For video/image stuff I found the ability for the LLMs to use ffmpeg and imagemagick to be quite fun.

I have it record a series of gifs or videos that I look over. If something looks off I'll dig into it, but I break down work into very very small chunks that are usually easily verifiable or don't require multiple steps.

Another thing I have in the general sdlc process is having it add enough logging to verify features are turned on, configured as we expected, and that becomes enough feedback for most of my features.

I've been mostly focusing on being able to replicate this across stacks greater than 3 projects so far (with the eventual goal of having an agent be able to orchestrate our complete infra stack, and this being a large component of a DR plan to rebuild).

None of this is really new for us, I'm just the most knowledgeable in my group in how the different products across teams glue together so I've been creating these rube goldbergs as a prototype, and then having it iterate on codifying the parts that don't need a constant LLM. We were blessed to have an engineer a decade ago build out tooling for local container automation that matches 95% of the deployed infra stack. That last 5% sucks when you fall into it, but that's always been a truth. I've added and expanded the tool over the years with making it act more like the deployed environment networking wise, but a lot of things don't end up working well in docker containers on M series macs when most of our complicated virtualization in our private cloud can't run on them yet...

This is where most of my productivity gains have come, I have a special harness I move from project to project now that does my testing orchestration, lots of my work day is setting up a prompt or two early and just letting them loop till they return evidence that the feature is working having gone through the big QA loop.

I've slowly been optimizing for token use through the stack and Claude ends up making very tight for loops for most of the process and keeping token count even lower. It's been nice. A lot of my toil at work is just gone.

Hey thanks for the note about the discord!

I have also been finding the MCP auth story to be really lacking was excited to see OAuth 2 support until I tried to get it to work at work and realized our idp implementation didn't support 2.1, and went into the spec and started wondering if anyone had a good experience yet. Luckily most of our environment can settle on a OAuth token env var standard until that's all in order.