Nash (@nash) 's Twitter Profile
Nash

@nash

Helping you use AI at work with @integryHQ DMs open. Ubuntu ngabantu.

ID: 20823

linkhttps://integry.io calendar_today24-11-2006 20:15:26

12,12K Tweet

3,3K Followers

2,2K Following

Nash (@nash) 's Twitter Profile Photo

Fascinating take: when you train LLMs with language, you’re technically putting the prism of human understanding in describing the universe. Which is one level removed from direct observation. Training only on perceptual data yield more novel insights for which we may not even

Nash (@nash) 's Twitter Profile Photo

The finder icon is weird because you wouldn’t expect the file browsing icon to be a face. But it’s very much the original Apple vibe. It’s a user whose face is being reflected off the screen.

Nash (@nash) 's Twitter Profile Photo

Modern day chips are “miniature skyscrapers” is the best description I’ve heard regarding the immense complexity in today’s “2um” chips

Nash (@nash) 's Twitter Profile Photo

I disagree, test-driven development is more important than ever with AI. If you’re generating code without tests, you’re gonna have a bad time. TDD ensures testable code, its baseline requirement for good code.

Nash (@nash) 's Twitter Profile Photo

New in iOSes 26.1, you get to pick between Clear and Tinted Good design is opinionated. Apple has always made design decisions on the user's behalf. This "option" is the design team giving up on making the hard choices and letting users figure this out.

New in iOSes 26.1, you get to pick between Clear and Tinted 

Good design is opinionated. Apple has always made design decisions on the user's behalf. This "option" is the design team giving up on making the hard choices and letting users figure this out.
Nash (@nash) 's Twitter Profile Photo

I’ve not been a fan of AI video, I think they give the uncanny valley and the ick. But with Coke making a major campaign ad using AI, I wonder what data do they have about regular users reacting to their AI video that makes good financial sense

Nash (@nash) 's Twitter Profile Photo

Kalashi is a great counter-example of the wisdom of the masses Or it’s an easily manipulated disinformation engine

Nash (@nash) 's Twitter Profile Photo

This is kind of insane. Anthropic started as a perhaps-alternate to OpenAI and now their APIs are so much better for agentic application development. If you’re a dev that’s been butting heads with gpt5, there’s so much less pain with Claude 4.5

Nash (@nash) 's Twitter Profile Photo

Spent 45m with a human on support chat to figure something out. They solved it and later sent an email saying, sorry they had to undo the solution because of a policy issue. Yeah, AI support can’t come fast enough.