Hey Community,
This is a series of concepts aimed at imagining mobile operating systems for the future - setting a vision for the agentic and AI-driven OSes and bridging our way towards it. Make sure to visit this thread to see a new concept unfold every Friday and share your thoughts and complementary ideas with the community to contribute to this aspirational vision for the future.
01.. Dear chaos, get off my albums!
02.. Contextual Actions
03.. Content Feed
04.. Next Friday!
Can’t wait to see what’s up here! 
Hello, it’s Friday which means we have our first concept of the series out!
I would love to have a good discussion with the community members on what they think about this feature, and any possible improvements that can be done.
I want an AI to plan my British Rail journey and buy split tickets at the cheapest possible price such that I don’t need to leave the train. That would be useful.
But, we are discussing photos here, and 99.9% of the photos I take these days are for sending to friends/family and thus I find clean up options/folder/management kinda useless (I have about 100000 photos since my first digital camera in 1999 and manually save them and hardly use them unless I want to show my children something interesting about my past).
http://camera-wiki.org/wiki/Fujifilm_MX-1200
I do find the automatic reviews via “AI” from times past via Google Photos useful to share with friends/family around the globe. It does remain a shame that WhatsApp/FB Messenger/Telegram/Signal/LINE/WeChat can’t process these “photo mosaics” and automatically send out to the relative contacts on their respective apps with the push of a button.
Now, that would be interesting.
CMF1_hacker 99.9% of the photos I take these days are for sending to friends/family and thus I find clean up options/folder/management kinda useless
Do you take multiple shots to make sure to get a right one? Clean up helps in keeping the best among them.
Uday Nah, after 100k+ clicks, it’s almost always one and done if the subject is willing 
vovan-v Yeah, that’s what AI is supposed to do, right? It’s supposed to make our lives easier, make it more efficient rather than playing gimmicks generating those uncanny images. It’s high time we start to automate things using LAMs, etc.
CMF1_hacker Ah that’s good, it’s quite common for others to have such duplicates in their albums, and hence thought this could be a great application of AI.
Hi! Uday here, commenting from my friend’s phone because I have to post the second one tomorrow 
It’s Friday and here is the second concept of the series - 02.. Contextual Actions! Make sure to read through “There’s more to it!” in the video, the real deal is in there!
Also, please share any complementary ideas or thoughts you might have!
Great work on this, Uday. Some stellar ideas for sure. Really like how you are trying to push the boundaries of what AI can do.
I would love to see a fully on-system AI one day that already has all the info pre-built into it and one that doesn’t require an internet connection. This will make people more comfortable with AI I believe. I don’t think it’s going to happen though.
Thank you for the appreciation, Rahil!
I am having a lot of fun working on this project, drafting a template for a range of exciting functionality AI could bring to the table. It feels great imagining a personalized on-device AI automating such time-taking stuff.
I think this functionality is quite achievable with the current-day AI capabilities or at least in the near future, but just that no one does it. Is it hardware limitations or financial implications that makes you think it’s not gonna happen? 
see you in the comments!
