jmogly

3 ms·

1.
by jmogly 2mo ago
He definitely is doing it for the love of the game. Different generation loves to work
2.
by jmogly 2mo ago
I watched my 75 year old human plumber (who sniped a guy his first day in vietnam), kneel on two 3 inch wide joists and drill a 4 inch diameter hole at a 89 degree angle through said joists. Where’s the robot that does all that? Above from
3.
by jmogly 3mo ago
or, just let people call things what they want, and focus on real problems instead. Even this comment as I’m typing feels like too much time committed to the subject.
4.
by jmogly 4mo ago
I got the feeling watching that last season of Stranger Things that the storyline and some of the dialogue felt AI generated. It sucked by the way.
5.
by jmogly 6mo ago
Yes I suppose this is the bare minimum, but isn’t that just a nasty way to go about things? What about responsibility and decency, do we just not do that anymore?
6.
by jmogly 6mo ago
How do you know that the llm is correctly translating the english queries to the verifiable primitives? It seems like it’s just pushing the problem to another layer?
7.
by jmogly 7mo ago
Ah yes great point, awesome article by the way —- thought provoking, shocking, really crazy stuff. Hopefully some good comes of it, godspeed.
8.
by jmogly 7mo ago
It is kind of a fundamental risk of IMDS, the guest vms often need some metadata about themselves, the host has it. A hardened, network gapped service running host side is acceptable, possibly the best solution. I think the issue is if your
9.
by jmogly 7mo ago
Mainly for getting managed-identity access tokens for Azure APIs. In AWS you can call it to get temporary credentials for the EC2’s attached IAM role. In both cases - you use IMDS to get tokens/creds for identity/access management
10.
by jmogly 7mo ago
This is insane, when you say azure OpenAI, do you mean like github copilot, microsoft copilot, hitting openai’s api, or some openai llm hosted on azure offering that you hit through azure? This is some real wild west crap!
11.
by jmogly 8mo ago
Haha, you can already see wheel reinventors in this thread starting to spin their reinvention wheels. Nice stuff, I run my agents in containers.
12.
by jmogly 8mo ago
OP never mentioned letting the agent run as him or use his secrets. All of the issues you mention can be solved by giving the agent it’s own set of secrets or using basic file permissions, which are table stakes. Back to the MCP debate, in
13.
by jmogly 8mo ago
Chat gpt is a great name though — you “chat” with the “GPT” so its self informing (even if you dont know what a GPT is), it’s 4 syllables that roll off the tongue well together. RSS, has no vowels, no information, and looks like an alphabet
14.
by jmogly 9mo ago
It would be like if I was expected to deliver A by the end of the quarter and instead I delivered A + B. The value gain from B was more than A. Your manager (and hopefully higher up the org) better know about B, or they will attack it as a
15.
by jmogly 9mo ago
Early in my career I would build something I thought was useful, deploy it, meet with people within the company to get people to start using it. A lot of effort for something that would have a positive impact. My manager would schedule a me
16.
by jmogly 10mo ago
Like it, a lot. I think the future of software is going to be unimaginably dynamic. Maybe apps will not have statically defined feature sets, they will adjust themselves around what the user wants and the data it has access to. I’m not enti
17.
by jmogly 11mo ago
Of course, step one is always critically think and evaluate for bad information. I think for research, I mainly use it for things that are testable/verifiable, for example I used it for a tricky proxy chain set up. I did try to use it
18.
by jmogly 11mo ago
I would say it varies from 0x to a modest 2x. It can help you write good code quickly, but, I only spent about 20-30% of my time writing code anyway before AI. It definitely makes debugging and research tasks much easier as well. I would co
19.
by jmogly 11mo ago
I’ve been using the following pattern since gpt3, the only other thing I have changed was added another parameter for schema for structured output. People really love to overcomplicate things. class AI(Protocol): def call_llm(prompt: str) -
20.
by jmogly 11mo ago
100% Gemini + pydantic you don’t need a wrapper library in 2025
21.
by jmogly 11mo ago
On one hand, I would like to say this could happen to anyone, on the other hand, what the F?? why are people passing around a dataset that contains child sexual abuse material??, and on another hand, I think this whole thing just reeks of t
22.
by jmogly 11mo ago
Eh I’ll take my 78, someone else can have the rest.
23.
by jmogly 1y ago
Lol! This is probably my sneaky number one productivity benefit from LLMs, I would never want to go back to writing shell scripts pre-llm. So many hours wasted debugging and deciphering stack overflow over the years, dropping &,$, [[]],
24.
by jmogly 1y ago
Some real honest and actionable advice here. I think the natural course for intelligent people that enjoy crafting things is very much in conflict with the real world. We care about the things we are building because we see them as an exten
25.
by jmogly 1y ago
Im also wondering how he is able to see calls to AI providers directly in the browser, client side api calls? Thats strange to me. Also how is he able to peer into the rag architectures? I don’t get that, maybe GpT4.1 allows unauthenticated
26.
by jmogly 1y ago
Python and Rust are probably the front runners
27.
by jmogly 1y ago
I hit it quite frequently with a particularly popular eks node instance type in us-east-1 (of course). I’m talking requesting like 5-6 instances, nothing crazy. Honestly, I wonder if ecs or fargate have the same issue.
28.
by jmogly 1y ago
Respect, sometimes we forget that our software is actually supposed to do stuff. Will say though you can have both, especially now with llms where you don’t have to trim each piece of wood yourself. Also, it’s not always about cutting corne
29.
by jmogly 1y ago
Effortless lying, scary in humans, scarier in machines?
30.
by jmogly 1y ago
I think dismissing social realities as not being part of that hard, unyielding reality is a mistake. Part of intelligence (maybe a different part) is being able to bend the social fabric to achieve desirable outcomes. The other thing is you

More ›