Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Anthropic Debuts Claude Design for Creating Prototypes, Pitch Decks, and Mockups
- Thread starter MacRumors
- Start date
- Sort by reaction score
Stitch by Google is better and Apple is light years behind on the software front. MacOS should have something like this built in. This and a chatbot.
Apple should have this built into MacOS but instead we get pathetic "Image Playground" and FreeForm with features from 2017. Apple really needs to step up the software aspect. Google built "AntiGravity", Stitch and a Gemini Chat bot all of which are better than Xcode, Freeform and Image Playground.
That’s what I don’t get. If there was a valuable business model in all this they wouldn’t be making it available to you and I.
Instead it would form the backbone of a never before seen business.
rp2011
macrumors 68040
Zip it dude! lol This is what I LOVE about AI at the moment. Just enjoy it while it's cheap. We can complain later when the prices go up.
Iridescent Caterpillar
macrumors newbie
Never seen a sheep beg so much to be taken to the slaughterhouse before.
uacd
macrumors 68000
I guess we are heading to the era where UI/UX designers would be no longer needed. They will either have less job to do, a lot of work with prompts or no job at all.
I hope this thing can at least generate something better than the slop minimalist UI designs that these same designers were making for more than 15 years already. What’s in the picture is not really “tasteless” but kinda boring. Can it generate a Y2K inspired website with funny fonts, shadows and backgrounds?
I hope this thing can at least generate something better than the slop minimalist UI designs that these same designers were making for more than 15 years already. What’s in the picture is not really “tasteless” but kinda boring. Can it generate a Y2K inspired website with funny fonts, shadows and backgrounds?
uacd
macrumors 68000
Indeed. Xcode is so ancient and bloated at this moment that the only reason why it is still “popular” is because Apple doesn’t give any other option to build and deploy for iOS. This is what actually those companies like Fortnite should be fighting for, not some useless alternative marketplaces. If anyone can build app on any device and even on the web, with AI or without, this would make things more accessible and in theory we would have more apps. Apple is probably afraid because they are not prepared and know their security system will get cracked very fast despite the notorious “sandboxing” here and there
All template apps have ever done to software design is make it derivative. Every app looks the same now and A.I. isn’t making things better.
I think these companies are going to run up against human boredom very quickly very soon. People grow tired of things increasingly quickly thanks to the speed of the internet.
Will Co
macrumors 6502a
Isn’t this the standard ensh*tification ploy we’ve seen elsewhere? Offer up something for nothing or at a price way less than competitors can do by burning a big pot of cash until those competitors have gone and you own the market. Then drive up profit, usually by cutting quality or simply putting up prices and everyone has to follow because there’s no choice.
timcorin
macrumors member
Nah man. Senior UX designer here. This is very cool for blue sky prototypes of ideas out the back of innovation workshops or for leadership to communicate an idea. But for genuine app design that needs strict ties to our design system components and how we need to lay out flows and documentation for development teams etc etc etc. It’s not something we would use anytime soon. But I use Claude pretty much hourly for work in a raft of other ways, including design reviews which is quite cool not to mention strategic planning and research synthesis.
Anthropic is internally using mythos for good fooding. This includes using reasoning traces of mythos to train next versions of opus and sonnet.
Apple are behind with models. However Mac sales are about to sky rocket with demand for Mac mini and studio shooting through the roof. They are the most cost effextive hardware for running local models. Cheap to buy and cheap to run due to low electricity consumption AND can be clustered.
This week I ordered a 48GB ram Mac mini pro with 20 gpu cores. An pc or Linux server with the same about of vram would cost 3-4 times as much but also have a 2000-3000 watt energy consumption. My Mac mini won’t be connected to a screen. I will communicate with it via text, calls and emails. In general it will preemptively do work for me whilst I am asleep or doing other stuff on my main machine 😎
By the end of the year my business will likely have a rack of 3-5 Mac minis with each serving as an ai employee, entering powered by local models.
Populus
macrumors 604
Each with 48GB of RAM? Or less? I’m still not sure how capable a Mac with 32GB would be with local models…
To run highly capable models for agentic/autonomous workflow, you need 48GB Mac Mini.
I will be using the Mac mini for coding. The model I will use is Qwen2.5-Coder-32B. That will be loaded into the vram for the GPU. Roughly 20GB. Then 8GB ram for MacOS, 8GB for AI context window and 8GB for various tools and applications.
You can run smaller models with 24GB or 36GB but they won’t be as capable for complex problems or long horizon tasks e.g Task, test, iterate loops over several hours.
For £1,870 that I paid for my Mac Mini, it’s a bargain. Lots of coding tools, including OpenAI and Anthropic, are lowering limits for paid plans. The era of subsidised inference is over. If I am to code with AI the way that I did before Feb 2026, I would now need to spend £300-700 per month. For that money I’m better off buying Mac minis and generate code 24/7 without rate limits or stringent daily or weekly quotas.
When the M5 Mac Mini is announced, it will sell out very quickly and have long lead times. It hands down the BEST and cheapest hardware for running local AI inference.
Populus
macrumors 604
I know, that’s why I’m highly interested in the M5 Mac mini, and I’m trying to gauge what’s the best value for me. Honestly for general use, I see the M5 Pro way overkill for me.
Because I’m not going to use it to code, but rather as a sort of RAG and file analyzer/organizer, I’m not sure I really need those 48GB. I’m not going to use it for agentic AI, or chatting to it, but rather as a tool for learning faster.
Combine this with the speed at which models are getting smaller (and now also KV with turboquant), and I think the 32GB should be useful in the long term.
But I’m not by all means asking people to tell me what I want to hear, but the truth. However, I’m not sure I’m willing to invest nearly 2.000€ on a Mac with an M5 Pro chip if I’m not going to make the most out of it.
ProbablyDylan
macrumors 68040
Claude, help me make my design language legible. Someone made everything transparent thinking it was a good idea.
32GB will be fine for your needs. Some people also approach a hybrid approach they includes local models and cloud based.
Basically you can give the agent on the mac mini a monthly €5 budget to tap into cloud models for planning or review and let local models do the work in between.
If after 6-12 months you feel you don't need it, all it and you will likely get 90% of your money back.
turbineseaplane
Suspended
If AI is going to "make everything" and can "consume everything" ..
Do any of us need to be in the flow of communication here?
Seems like AI's should just communicate directly to each other.
The humans are an unnecessary middle man.
Whether what the AIs are "doing" is going to ultimately be of any value is somewhat of a question here, particularly if humans are removed from any part of the chain.
Do any of us need to be in the flow of communication here?
Seems like AI's should just communicate directly to each other.
The humans are an unnecessary middle man.
Whether what the AIs are "doing" is going to ultimately be of any value is somewhat of a question here, particularly if humans are removed from any part of the chain.
AI's understanding is limited to information that it can scrape on the web.
Only 5-10% of all human Knowledge and behaviours have been recorded online. Lots of it is private or has never been written down.
You can tee me out for steak and beer and get firer hand information that ai doesn't know yet. There your asymmetric information advantage.
turbineseaplane
Suspended
That seems really low to me.
I'm curious, where are you getting that estimate from?
I have worked at multinational businesses across several industries and run a business myself. I know lots of information that chatgpt or Claude Web does not because the internal information has never been made public.
I also once listened to an entrepreneur in california who started running a business that targeted pregnant women in an Asian country.
After a year he learned that there is only so much that he could find out sitting at his laptop in san fransisco. He ended up moving to the Asian county because women there don't post their lives on the web for you to learn about them. You simply have to be on the ground and in the community.
rp2011
macrumors 68040
