Chase Kunz

Well written and exactly how I feel about it as well.

this has to be ragebait i refuse to believe these people are real

working on adding agility to kunzaria here’s a first wip look

sometimes you just need to message the app review team in case there was a mistake... my app was automatically rejected for crashing, but when asking for logs they decided to do an actual review and... it passed? i still worry there is something that could be crashing the app, but i don't have crashlytics (that's on me), so i guess i'll wait for a complaint

I think OpenAI took "pacing the frontier" too seriously...

oh no... this really was the codex-y vibe. this is... not great.

is codex-y in the room with us?

this is good, but not certainly not enough for codex-y thursday what else we got?

please be codex time please be codex time

just downloaded t3 code to give it a try

if I use a reset will the reset company release something today?

Codex on Duo needs to show images/artifacts on the right side like Claude does. Including the >Viewed an image GPT does in-line. It’s annoying to click each one and have to X out of it just to click the next and the next etc

Soo… the ChatGPT Duo app doesn’t work with an embedded PiP in opened portrait mode. Keyboard tries to open and then it spazzes out and then the sidebar opens. Send help @Dimillian

my dot created an .mp4 but now has no idea how to send it to me/embed it in chat despite doing it hundreds of times before...

somehow ended back on dia ran into issues nearly every hour on zen

Not sure why Claude has this attachment limit... it's really annoying to be working on mobile and then you need to do a workaround to be able to see results to continue working.

I've been subscribed on this Plus account for ChatGPT since January 2024. I still use a Premium Business account and get good value out of that, but I don't have enough usage that requires multiple accounts anymore (and it's not because the usage limits are out of this world). Nearly everything UI related that I created with Astra/Sol required an extra pass by Opus if I wanted good polish (or spend 5 more years prompting GPT); however, UI related things that I created with Opus did not need a pass from GPT (or non-UI related things for that matter). For example: I wanted to round off my left-nav sidebar for a project. I thought it was a pretty simple request. I even provided a partial example. When I asked Opus, it did it 100% perfect the first time. So on a different project when I decided to make the same design decision, I was already working with GPT (likely 6.1 Sol High or xHigh I can't remember), and I sent the same prompt with same reference I sent Opus. First, it didn't hide the scrollbar that cut into the rounded edges, then it also rounded the bottom of the page and added an additional border. You can think, "oh, well why didn't you just tell GPT to due that in the first place?" The answer is because Opus previously had done it without me needing to, so I'm likely to use the model that provides better results while also requiring less literal instructions for every little thing. Some things should be a given.