This is incredibly useful. I love the part where it shows who made the change. Plus, there's no need to even communicate the changes through conversations because anyone can immediately modify the way something looks and add an annotation to explain their reasoning.
You've made many points that I've thought about since Framer became popular with certain content creators bragging that they made 100k dollars in one year just selling templates. Framer and webflow template market is saturated by this kind of nonsense designs. There are many “dribbble” templates, many AI generated, there is no passion, no creativity, everything is made to look “pretty”. But it is zero functional. It's just people wanting to make money.
Most of that social media bragging is fake, don't worry about it. Functionality prevails in real life scenarios. Some clients fall for "just pretty" and quickly realise their mistake.
We made a proof of work for the edu platform - the bot gave about 8000 project feedback comments in v1. Now we're adding more features to V2 and making B2B :)
Not totally related, but you mentioning the AI being able to scan your text even if it's an image reminded me: we've had OCR and spell check for decades now - why does AI have such trouble with text? Those two things seem like they should prevent it from hallucinating random scribbles in place of text. I don't get it.
Yes, it's an oversimplification, we use OCR and then feed the result to the AI in real life. But sometimes in a video you take mental shortcuts. Which reminds me, I knew a startup where you took photos of store receipts and it "scanned them" back in 2011-2012. The reality was it was sending the images to 40 guys in India who typed the text manually working in shifts 24/7. True OCR ;)
@@MalewiczHype Oh, yeah... I wasn't really talking about the video at all... just the fact that AI can't seem to get text right even though we've had software that can identify characters (letters) and can check spelling (words) for years... OCR was developed in 1947! But ask a generative AI to make something with text in it and it's garbled nonsense. Like, it can tell you in text that it's made your image for you, but it can't make an image telling you it made the image - when it's effectively the same thing.
@@BoredToDeafRecords It all depends on how AI models have been trained. Most of the available models used a huge number of pictures to learn, so they can accurately recreate most of it. AI models draw text and numbers in images, they don't write it. Because they don't understand it the way we do. We see words, AI sees lines and circles. If you use hundreds of pictures with the alphabet on them, the model will learn it and you can recreate it. But it won't be able to create words with it.
@@hotshot6674 You're totally ignoring the OCR angle. OCR has been a thing for longer than we have been alive (unless you were born before 1947). OCR + spellcheck (+ grammar check) means hallucinated text shouldn't have ever been a thing. That it "draws" text is irrelevant... and because OCR is a thing - just like it knows what a human looks like - it knows what a J looks like, otherwise OCR wouldn't work. Hallucinated text is the giveaway that Generative AI isn't working the way we've been told, because that's the one thing it should have already been good at (all the guardrails were already in place).
This is incredibly useful. I love the part where it shows who made the change. Plus, there's no need to even communicate the changes through conversations because anyone can immediately modify the way something looks and add an annotation to explain their reasoning.
This looks promising, especially for the copywriting, people who struggle to articulate in English. Eagerly waiting.
There's also other features I cannot share now - I am extremely excited! :)
It's crap, don't even bother.
Looking forward to this. Also stoked to see one of my LP designs featured in a video. 😀
We cherry picked the very best :)
You've made many points that I've thought about since Framer became popular with certain content creators bragging that they made 100k dollars in one year just selling templates.
Framer and webflow template market is saturated by this kind of nonsense designs.
There are many “dribbble” templates, many AI generated, there is no passion, no creativity, everything is made to look “pretty”. But it is zero functional.
It's just people wanting to make money.
Most of that social media bragging is fake, don't worry about it. Functionality prevails in real life scenarios. Some clients fall for "just pretty" and quickly realise their mistake.
Interesting idea. I wonder how it will look in practice :) I'm keeping my fingers crossed! 🔥🤞
this will also be great for freelancers to test their work before delivering it to the client. 🫡
So exciting! Thanks for sharing valuable information, too :)
I really like this idea, just subscribed to it 💝
This seems like a really cool company and developer approved 😎
Can't wait to start letting people in. What we have here behind the scenes is 🔥🔥
@@MalewiczHype Everything you do is 🔥🔥 I’m still blow away by what you have done at the Square Black Academy 🙏🙏 It is a designer dream come true!
Great design content
New SaaS project?? Looking forward to it! Seems promising
We made a proof of work for the edu platform - the bot gave about 8000 project feedback comments in v1. Now we're adding more features to V2 and making B2B :)
Cool idea to connect it to the platform 🎉!
Parts of it we'll reuse elsewhere too - including our teaching materials :)
do you have something that apply to web design, colors,psihology, fonts,layout etc ? any ebook that cover all websites and not mobile aps ?
I have a full 3-part web design course. I believe its my best edu work. 7000 students on it seem to agree :)
@@MalewiczHype sorry for my previous typo. If is not so expensive , can i have a link ? Thank you
This is a big announcement 🎉
Not totally related, but you mentioning the AI being able to scan your text even if it's an image reminded me: we've had OCR and spell check for decades now - why does AI have such trouble with text? Those two things seem like they should prevent it from hallucinating random scribbles in place of text. I don't get it.
Yes, it's an oversimplification, we use OCR and then feed the result to the AI in real life. But sometimes in a video you take mental shortcuts.
Which reminds me, I knew a startup where you took photos of store receipts and it "scanned them" back in 2011-2012. The reality was it was sending the images to 40 guys in India who typed the text manually working in shifts 24/7. True OCR ;)
@@MalewiczHype Oh, yeah... I wasn't really talking about the video at all... just the fact that AI can't seem to get text right even though we've had software that can identify characters (letters) and can check spelling (words) for years... OCR was developed in 1947! But ask a generative AI to make something with text in it and it's garbled nonsense. Like, it can tell you in text that it's made your image for you, but it can't make an image telling you it made the image - when it's effectively the same thing.
@@BoredToDeafRecords It all depends on how AI models have been trained. Most of the available models used a huge number of pictures to learn, so they can accurately recreate most of it. AI models draw text and numbers in images, they don't write it. Because they don't understand it the way we do. We see words, AI sees lines and circles. If you use hundreds of pictures with the alphabet on them, the model will learn it and you can recreate it. But it won't be able to create words with it.
@@hotshot6674 You're totally ignoring the OCR angle. OCR has been a thing for longer than we have been alive (unless you were born before 1947). OCR + spellcheck (+ grammar check) means hallucinated text shouldn't have ever been a thing. That it "draws" text is irrelevant... and because OCR is a thing - just like it knows what a human looks like - it knows what a J looks like, otherwise OCR wouldn't work. Hallucinated text is the giveaway that Generative AI isn't working the way we've been told, because that's the one thing it should have already been good at (all the guardrails were already in place).
@@BoredToDeafRecords And what exactly has OCR to do with machine learning? It seems you don't understand the concept of it.
when subscribing, I get redirected to a mailchimp page, is that normal?
For some people it does that for some reason (mailchimp related) but it does subscribe you. For most people it works ok.
What does the AI say about its landing page? Seems for me like it probably would have roasted it hard. There is so much off
this lp is temporary, hacked in 30mins so we can put full focus on the platform. For now this works, people register.
Amazing……
Is that AI version of you in the thumbnail?😃 Which one did you use?
In one thumbnail - yes. We used our own model that works on Flux.
@@MalewiczHype nice, i havent tried flux yet, but im using stable diffusion comfyUI locally on pc to do the same
Figma Dev Mode
too expensive
First to comment, Mike ❤
It seems someone else got here first
@@MalewiczHype Oh! I missed that 😅 Cool stuff you are building, Mike ❤️
Yupp full build mode!
first :)
Congrats, for being the first here's a firm handshake: 🤝