What "if you know what you're doing" actually looks like: The model already has incredible general smarts… but what it doesn't have is what you specifically know about your field (it may know a LOT, but an expert can usually push it past what it knows pretty fast), or the
@mattshumer_
-
Massive capability gains from models if you know how
By
–
I'm not sure if it'll last (I go back and forth), but today, there's massive capability gains you can pull out of the models if you know what you're doing.
-
Matt Shumer says Fable solved computer use
By
–
Also, I don't know if anyone noticed, but Fable ~solved computer use.
-
Testing complex APIs with goal mode and sub-agents
By
–
Try 5.5 on xhigh, scope the prompt well, ask it to write tests first, run /goal mode till a) all tests are green and b) a sub-agent with a clean context window can test the API and it works perfectly Overkill for 99% of APIs but maybe yours is extremely complex?
-
Use loop prompt with Claude to perfect impossible tasks
By
–
I don’t usually get excited about Claude models at all. Look at my track record. When it comes back, try it on a task you feel is completely out of reach for modern AI. Use a /loop prompt that provides a way to verify the result, and tell it not to stop until it’s perfect.
-
Seedance 2.0, Hyperframes, Remotion with Claude for programmatic videos
By
–
Seedance 2.0 for pixel video generation, Hyperframes and Remotion with Claude models for programmatic videos.
-
Fable’s magic lies in tasks beyond other models’ capabilities
By
–
Seeing so many people saying Fable wasn’t that much better than other models. Sure, if you’re just doing normal frontend/backend work that Opus/GPT 5.5 can already do, it’s going to feel basically the same. The magic comes when you’re asking it to do things other models can’t.
-
Opus not good enough for 3D and verification
By
–
Opus is nowhere near good enough at 3D/verification. I’ve done some Opus ultracode runs and it’s passable, but not great!
-
Claude Code loop prompt and references for quality bar
By
–
Yep. Claude Code. Got this one in particular working really well with a loop prompt and a bunch of references to serve as the quality bar.