What would you like to see in a local LLM app beyond just a chat interface? Im seriously taking all ideas into consideration for my project, what would make an AI genuinely helpful to you? Thank you for taking the time! Means a lot
What would you like to see in a local LLM app beyond just a chat interface? Im seriously taking all ideas into consideration for my project, what would make an AI genuinely helpful to you?
Hi everyone! I’m a student developer interested in AI, and I recently released my first iOS app, Orion Personal AI.
Orion is a fully local LLM for iPhone. It’s built with Swift/SwiftUI and runs entirely on device using Dolphin 3.0 (Llama 3.2). No account is required, and no conversation or personal data ever leaves your device.
Rather than trying to compete with cloud models like ChatGPT, my goal was different. I wanted to create a privacy first, offline AI assistant that anyone could use without downloading models, configuring settings, or dealing with technical setup. Just install the app, complete the onboarding, and start chatting.
I spent a lot of time designing the onboarding experience because I wanted Orion to feel more like your own personal assistant than just another chat interface. The onboarding is meant to be part of the experience.
The initial launch can take a little while, especially on older iPhones, since the model has to load into memory. I’ve tested it on older generation iPhones, and while it’s naturally slower than newer devices, it still works well.
One design choice I made intentionally was not to build a generic interface for loading arbitrary local models. There are already great apps for that. Orion is meant to provide a polished, zero setup experience centered around one carefully chosen offline model.
It also works on iPhones that don’t support Apple Intelligence, so older devices can still have a capable offline AI assistant.
I’d genuinely appreciate any feedback.
Thanks for taking the time to check it out!