jstaley
@TheJohnnyStaley
Followers
14
Following
60
Media
6
Statuses
39
when you forget to ask claude not to include your api keys in your public commit
0
0
5
i don't think it's intelligence that's limiting people. what i observe is that those who are not in tech circles are woefully unaware of the capabilities of current models and the software ecosystems that surround them. e.g. Claude Code, Codex, etc.
most people just aren't smart enough to quickly get an intuition for what LLMs are and are not good at, nor do they have the language skills necessary to really squeeze them for great performance
0
0
0
my UI for llm model selection. it actually let's you browse models from providers prior to entering an API key. it sounds like a small detail, but the commitment associated with setting up a key is not insignificant for most users
1
0
4
i agree. gemini 3 flash just solved a bug for me that gemini 3 pro (high) was struggling with.
so wait let me get this straight gemini 3 flash has RL improvements that have not even made it to pro yet? absolutely ridiculous. we have been testing flash & its performance has been astonishingly good. combined with speed & price, it’s a breakthrough for how quickly & widely
0
0
0
update: i have been collecting llm providers like infinity stones which has been even more satisfying. it would be fun to have the models compete in some way
idk what it is, but I find this list style very visually appealing. i think a hue slider would be a fun touch though. if there were a 5th provider that deserved to be on this list, who would it be?
0
0
0
idk what it is, but I find this list style very visually appealing. i think a hue slider would be a fun touch though. if there were a 5th provider that deserved to be on this list, who would it be?
0
0
5
i switched from hard-coded LLM API calls in my app to LiteLLM. there are an ungodly amount of providers/models that I don't have to maintain now. seems like this will be a great time saver
0
0
4
give it some time and i think people will start to realize Google kinda cooked with this. i've had great results so far
0
0
1
this is how West World happens
This is the coolest thing I seen about LLMs in long time. This guy is trying to train a model with only data from 1800s creating an llm that behaves like that society... So damn cool https://t.co/NprF9DDRtV
0
0
0
for anybody who wants to read/consume dense materials like this I highly recommend NotebookLM. you can make fairly entertaining podcasts from the material one chapter at a time. game changer tbh
4 months later, DDIA complete. All 12 chapters. Wonder how many people who post about this book have read the whole thing? Parting thoughts: - I see why it's so well regarded. Incredible breadth, well-researched, and many topics backed with examples and diagrams. - Writing v1
0
0
0
i think it comes more from fear than stupidity. people see others succeeding in spaces they've spent their entire lives studying. nobody wants to believe they've wasted their time
ai is stupid if you are stupid. thats why only stupid people hate ai.
0
0
1
Spam calls are getting out of hand. Someone please develop an app where I can provide my number, and any time an unrecognizable phone number calls me, an AI voice trolls the person, and I can listen to the recordings afterward.
44
4
73
Apple might be the only tech company still doing this. whenever they release a product, its hard to forget and the message surrounding it is clear. with Google, nobody has any clue that half of their tools exist
companies like google need a chief product officer or something. it would help consolidate & remove confusion as well as duplication of efforts. right now everyone is shipping so fast because that’s the “in” thing to do & will likely get you promoted but a ton of it feels
0
0
0
I've been using Opus 4.5 in Google Antigravity and this seems mostly true. when there are errors, it's almost always due to my failing to provide precise guidelines. agency and communication skills are the only moats left
Opus 4.5 + Claude Code is 95% AGI. Yes, you have to set things up for success. But once you do it can perform any knowledge work task at the level of a competent human. This did not exist even 3 months ago.
0
0
0
a pm is a prompt engineer for humans
i have never heard a simple clear one line explanation of what a product manager actually does which is kind of hilarious. what is it?
0
0
0
People are gonna judge you no matter what. You might as well be judged for trying and failing than not trying at all.
225
505
4K