Hello AppleVis community!
I wanted to reach out and mention that Perspective Intelligence, my on device AI app that uses Apple Intelligence is now on version 1.4. This version of the app adds new vison based features that can read text and describe objects in real time. The app can also use on device processing to describe a picture that has been taken.
I'd really love to hear your feedback on these features, and how we can make it better. I believe that we have the potential to build something incredible with this app especially if we do it as a community. I'll Monitor the thread here, and you can also send me an email at feedback@perspectiveintelligence.app. The website for the app can also be found at https://perspectiveintelligence.app
I really hope the app is useful to everyone, and I look forward to hearing your thoughts.
Comments
live AI mode and share screen features
I hope it will have live AI mode and the share screen features
Ability to store notes
I will soon be upgrading to a new phone. So would like this feature a lot.
Paid Features
- Is not even an app intended to be used exclusively by blind and visually-impaired people, and is used by sighted people as well, meaning it doesn't necessarily have a tiny user base and shouldn't be promoted on this baseless claim
- Does not offer any functions that no other accessible alternative can fulfill
- Offers functions found in other free apps, as part of the premium subscription
So the developer should considerMy apologies if I missed your email.
Hello, I want to respond to your thoughts here.
It is my goal to provide the best user experience for the users of Perspective Intelligence. I went and looked up the apps you mentioned, and one of them described the use of Apple Foundation Models incorrectly saying it uses Apple Cloud Compute off device, which it doesn't. Also, many of these apps look great, until you realize that you can't really complete tasks with them. Perspective Intelligence can do what those apps do, and more, because we add AI tools for getting things done.
Something that you may not realize is that the only features that are paid are tools that users can use to complete tasks. All basic AI chat functions are free, which include Chat, Image Descriptions, vision features and similar. THe only reason audio transcription is paid is because it is thought that it will be used in transcribing meetings or taking notes. I am looking to add features for deafblind users that will use similar technology, and those features will be free.
The goal of this app is to elevate on device AI for everyone, and I think we are doing that, but Let's not come on a website and criticize as has been done here. Let's be constructive and work with us. There are plenty of ways to get in contact. We have a WhatsApp Community, a Facebook group, and our email. We are more than happy to communicate. So, please use all mechanisms to get in touch with us if you have an issue with the app.
Will we change the payment structure of the app? No. We give a lot of features away for free, and we are continuously making more that we don't charge for. The biggest hurtle for us has been to build support for older phones. This has kept us from adding new features so we can make sure to get it right for everyone.
Once we have that then we will start adding more features. I'd like to add some really cool sruff so I hope everyone will be supportive and let us know what works, and what doesn't.
I would also like to mention that we all are in the same boat. We are blind developers. I have worked in the blind community for years, and was a tech instructor for a while, so I have a perspective on creating apps for the blind that many others do not have. That is why I hope this app is useful for everyone, and if it isn't, then we need to make it so and that can only happen if we all contribute.
Thank you.
Yeah, tell me to contact you and then don't reply to my e-mails?
Sorry, I didn't quite get the first couple of points regarding Apple Cloud Compute and task completion. Thing is, the reality may be more of the exact opposite of what you claim, because Perspective Intelligence requires a subscription to fulfill many of the functions found in other apps for free. I can't even browse HuggingFace repos to download custom models. The app downloads some mysterious model, which may not be what I like, so I didn't bother to actually download it and find out. Hiding the model name doesn't make you more professional. Displaying it makes you more transparent. Offering more options - perhaps a curated list - and then letting users search for HuggingFace repos and download GGUF or MLX models makes you more flexible. Many free apps have all those features, and may even let users import models from files or download MLX Audio models.
mac version
This app sounds awesome!
Are there plans for a Mac version, sorry in advance, if one already exists . I have an m1 air and could see myself using this for image discriptions on the web for example
An Update
Here's an update. Not from the app, nor from the dev, as he has not replied to my previous post, but from me. Who knows, perhaps the developer has been busy working on new paid features we can access for free in other apps. He could even use the opportunity to his advantage because I had certain points to be corrected there, and now I will have to correct myself. The developer tends to either ignore or somehow miss my posts and messages though, so this is not so surprising. Anyway, let's move on. The app does not hide the name of the model you download altogether, but you have to actually double-tap it to view the available options, which also means the developer does provide a curated list depending on the device model, which was another thing I suggested in my previous post. Still, chat and vision models are totally separate, and you can't have text-only conversations with models shown as vision models unless you first switch to vision and then take a photo to have the model describe it so that the option to switch to chat shows up. What's worse and still more important, is the fact that you can't search for and download custom models yourself. So this "We do and set up everything for you on your behalf but don't let you pick and customize anything to your liking." attitude is highly problematic, because many developers exploit this as a pretext to make their apps or certain functions found in other free apps paid, and they even promote their apps as lifesavers that offer essential features that no other app has.
About OnDevice AI
If you have been following on-device AI developments on iOS, it is pretty well known that Apple's foundation models are not the best small language models that a powerful phone like the iPhone 15 Pro can actually run. Maybe for people who are less curious and just want things to be as simple as possible, this is the way to go.
But for those who are curious enough to search for keywords like local AI or private AI in the App Store, it only takes a bit of time to see how much more power and how many possibilities your high-end Apple device actually has. You would be surprised at the performance level of open-source models found on Hugging Face when they are properly leveraged—apps like Enclave, Locally AI, OnDevice AI, Lekh AI, or Esper AI are great examples. Most of these are either totally free or offer a lifetime purchase for a small, justified cost rather than an expensive annual subscription. They are very accessible and for everyone, not just a specific group. I guess everything has its target audience, but it is surprising to see such a high price for models that are essentially the basic, entry-level options. Still, all the best to the developers of this app.
NoemaAI is one such example.