I like how @Muse does bg research on the topics and proactively reminds me on the future actions or updates. Maybe there should be a space somewhere I can see all the bg research it is doing on all the topics, so that I can just toggle pause it or turn it off. CC: @alexcornell

I almost never hit the 5-hour limits before Astra, even on Sol 5.6 High. After Astra, I’m hitting usage limits even on Sol Light in fewer than 5 prompts. Guess the days of cheap tokens are over. Good thing Claude hasn’t reduced its already-low usage limits.

I've become a QA prompt monkey. The core of what I envision gets built in first 2-3 prompts. Remaining hours I just keep prompting about quality issues, wrong icons, bad spacing, wrong typestyle, wrong states, its endless loop, this is even after having design system & figma MCP.

It’s gotten to the point where even using the internet feels like I’m burning tokens and about to hit my weekly limit.

Codex app experience feels much better for building web apps than using terminal. The in-app browser, queue management, div annotation, tighter GitHub integration, etc. It does burn through tokens fast though, hit my usage limits for first time in just a couple of hrs.

Created a fast follow, adjusted the shader container and wave position, SiriAI will appearing from dynamic island. Apple is most likely using a custom glass shader for the Siri container, I tried encapsulating the container shape with .glassEffect() but it didn’t look good.

Attempted to replicate new Siri AI animation from #WWDC with Metal Shaders. V1 is using Codex and V2 update is by Fable.