For gpt-5.5 I build up that session context into a markdown file, and then I start a new session and give it the markdown file with the instructions for what to do.
I'm guessing this works better because it can always go back and re-analyze the saved context.
Yes, I've just come to the end of implementing all the planning I did while Fable was available. And nothing now comes close to creating plans that could be coded and just worked like it did.
On a large C codebase, Claude hallucinates constantly, and GPT 5.5 gets there are with a lot of help, but still gets things wrong.
> It's also hard to brew coffee if you tend to use off-the-boil. The best you'll get is about 93 C.
That sounds ideal for off-the-boil coffee brewing? At sea level I (and all the speciality coffee shops round here) aim for 91C, and I'll drop that to 88-89C for medium roast and lower if it looks on the dark side. Brew methods: Aeropress and cafetiere.
Let's say you're thrown into a website you've never worked on before and asked to fix a styling problem. You can look in the browser tools, but the website will only be running the compiled production version, and if the team knows what they're doing there won't be source maps available.
So you've now found selectors in DevTools that you think are causing the problem, and you want to find them in the source code. In the case of many projects, that means searching through hundreds of small CSS files.
That's why you grep selectors, and where the pain comes. You have to start with the most specific rules that you found in DevTools, then start deleting parts from them until you find a non-nested rule that's in the source, yet still specific enough that you haven't got hundreds of matches to go through.
It would be great if something like ast-grep could take a CSS rule copied from DevTools and search for nested CSS that would compile to match it.
> Basically, I need Claude Code with a proper review phase built in. I need it to slow-the-fuck-down and work with me more closely instead of shooting mountains of text at me and making me jam on the escape key over and over (and shout WTF I didn't ask for that!) at least twice a day.
It sounds like you want Codex (for the second part)
> The truth is, business opportunities are rarely eternal, usually they are just an opportunity to make money within a short window of time, such as a decade or two. Sometimes even shorter than that, perhaps even only a year or two.
Agreed, and Adam and Steve made a life-changing amount of money from Refactoring UI and then Tailwind UI. That's a great outcome on its own.
FWIW I found Practical UI [1] a more actionable book than Refactoring UI. Both are similar but I found it covered the material in a more accessible way.
> Now, basically every new "AI" feature feels like a hack on top of yet another LLM.
LLM user here with no experience of ML besides fine-tuning existing models for image classification.
What are the exciting AI fields outside of LLMs? Are there pending breakthroughs that could change the field? Does it look like LLMs are a local maxima and other approaches will win through - even just for other areas?
Personally I'm looking forward to someone solving 3D model generation as I suck at CAD but would 3D print stuff if I didn't have to draw it. And better image segmentation/classification models. There's gotta be other stuff that LLMs aren't the answer to?
I'm guessing this works better because it can always go back and re-analyze the saved context.