I’m developing an AI email app right now, and really wanted to use on-device Foundation Models for privacy and cost reasons but they are just not good enough. Anything that gets Apple closer to acceptable LLM performance will be a huge boon to devs and users and enable better apps on multiple axes
@myke.social talking about broken Spotlight, Clipboard History has been completely broken for weeks on my Mac and nothing I’ve tried has fixed it. Zero history showing. Ugh.
Pretty clear from how many gigantic scientific strides came about by accident while trying to accomplish or study something else
@snell.zone @myke.social Surely the Ultra 3 requires a cell plan for cost saving. On iPhone, can only use when you don't have signal or wifi, which minimizes usage. If Ultra w/o cell can use satellite messaging everywhere (because never have signal), usage and Apple's costs go way, way up.
I don’t think OP is new slang. I first heard it in university in 2007 from a roommate who played Counterstrike.
For makers of products that seem to purposefully interfere with password managers working properly on your login screens... why?
In what world does this result in any behaviors that are better for security?
That’s cute 🫣
Yeah I mean, how could you not?
Sleep stages report from Apple Watch showing a wake period from 4:00-4:30am during an otherwise restful sleep
Nothing like waking up in the middle of the night groggily worrying about who will take care of us when we’re old and senile.
Then remembering how much progress has been made in humanoid robots and that we’ll definitely have an attentive robot nurse by then and we don’t have to worry about it 😂😴
It’s very impressive. Exceptional along dimensions that image gen models have traditionally struggled very hard with.
Not surprising. I feel like current RAG pipelines are still naive in a sense, too. I suspect going into the future there will be much more rigorous pre-processing pipelines to extract knowledge and information prior to embedding
That was a wild decision to not even mention it at all.
It was an incredible drive, but the rules are the rules and the team failed. We’ll never know if he only stayed ahead of Lewis because his car was too light.
Especially since France helping America out directly led to them going bankrupt
It depends why you're coding. If you are doing it for the joy of making things, then ease up on the AI. If you're doing it to accomplish a goal then go full steam.
This can vary from project to project (or day to day, even)
Screenshot showing Mac Studio build for $14,099 on Apple.com
Brb, going to look in the couch cushions for $14k USD to get half a terabyte of RAM and be able to run 600B parameter LLMs locally!
This season of Severance is killing it. What a great episode last night, wife and I just spent an entire hour-long walk talking through the implications and questions.
Great work @benstiller.redhour.com!
@liss.bsky.social excellent vamp while waiting to skip YouTube ad, but have you tried YouTube Premium? Totally worth the money, once you experience it, you can't ever go back.
And yes, I know adblockers, etc. But that is harder when watching a video on TV or on phone/tablet.
Ugh, so annoying. And often it’s because I want to check out the pricing page to look at maybe upgrading, or send it as a recommendation to a friend. This behaviour stands in the way of me evangelizing their software
I’ve been experimenting with different video models and I have to say that Kling is my favourite so far by quite a margin
Very proud of the product video I made for my AI email tool, Delegate! It was my first time diving into After Effects and motion graphics, and I enjoyed it quite a bit. Being able to work on diverse projects is my favourite thing about indie building.
Indies being told they have to post and comment like crazy no matter what, otherwise they aren’t “building in public,” was a huge disservice and contributed to a glut of noise that doesn’t really help anyone.
Contribute, whether a post or comment, when you can add *real* value to others. That’s it
I've been making a product video in After Effects (lots of fun, highly recommend) and it's crazy how quickly my opinion of my M1 Pro with 32GB of RAM has gone from "this computer will last me another 5 years" to "this is literal garbage and I want to spend $4000 on a new computer"
Such a thing as being too concise. They could spare a few more words there!
Agree. I think many actually flip the causality of the work <-> meaning connection.
Some people are lucky to do what they love, but many are forced to find meaning in their work otherwise it would be unbearable.
If you wouldn’t keep doing it after winning a lottery, then it’s not a calling.
Solid joke 😂
Dogs are maybe the best and most pure “good” in this world. So much upside and love for such little “cost”
I forked off the Twilio-OpenAI demo. Maybe I should take a look at the console example they have to make sure I’m on-side.
Has anyone worked extensively with the OpenAI Realtime API?
I’m finding it’s struggling very hard to understand my voice compared to the Advanced Voice Mode, and I’m hoping it’s due to audio handling that can be tuned and not just that it’s very bad.
It’s Schrödinger’s Entrepreneur
If 2024 taught me anything, it’s that you can eliminate churn by simply not giving customers an option to end their subscription 😂
