Infinity Vaultz

Private Offline AI, Secure Data Storage, and Safe Browser

Privacy

Artificial Intelligence

Security

I've spent the last 20 years building my own software companies from the ground up. Throughout that time we've seen privacy go out the window, tracking through the roof, and we now share our most intimate thoughts to chatbots made by large AI companies. We all have things we want to keep private whether it is what we are browsing, photo/video/files we keep, or the thoughts we seek answers for. It is not just about being sneaky or concealing, it is about having a sense of security and privacy in our daily lives. Over the years, I used various photo vault apps. Some were decent, some terrible, some outright unsafe, and I knew I could create something considerably better. We deserved better and so I built Infinity Vaultz as a passion side project to fill that gap. I didn't build it for money (there is a Pro tier, but most features are free and it exists to offset the costs of running this as a side project), I built it for the community that cares deeply about privacy and security (largely other developers). Version 1.X which has been live the last year fulfilled my goals on the privacy and security side of things. I researched 50+ other photo vault apps and looked for what features seemed most valuable, which issues other apps failed at, and privacy holes that were not filled. This resulted in a product that: a) Multi-layer security including pin, password, pattern, biometric, geolocation, and 2FA authentication with decoy support and break-in records support. b) Has a full feature private browser with ad/tracker blockers and easy ways to download/record photos and videos from the web. c) Stores photos/video/audio/files with AES-256 encryption using your own custom encryption keys in one or more vaults (both offline and/or your OWN Dropbox/Google Drive fully encrypted). d) Has a private camera that stores photos/video directly in the app e) Various other capabilities like Notes, Reminders, Wallet, Passwords, etc. f) Built-in ML tools to work against your data. Feedback has been overwhelmingly positive thus far from those using it. With that said, as good as it is, 1.X "is just another photo vault app". So I decided to think about what would make it truly unique and groundbreaking for 2.X. The time has finally come where mobile devices are powerful enough to run AI inference locally. No, they can't run trillion parameter models BUT they can run models under 8 billion parameters well and run stable diffusion image generation models. So, I thought to myself, how can I create an offline AI chatbot that isn't just a silly toy that answers "why is the sky blue?". A good AI chatbot would require: a) Ability to pick the model you want. I chose a series of MLX open source models on Hugging Face like Qwen, Gemma, Llama, etc. as well as the ability to use your own macOS Ollama server or your API based connections to OpenAI, Anthropic, and Gemini (all optional; the on-device models need no connection at all). b) Image generation. I chose to integrate various Stable Diffusion models. c) Ability to customize the models through templated Agents with pre-set system prompts, temperature/topK/etc. and ability to create completely custom agents yourself. d) Access to tools like web search, reading urls, and most importantly chatting about your own data in the app. e) Memory capabilities f) Ability to manage your chat history g) Handle long context windows gracefully with limited memory and not crash h) MOST IMPORTANTLY, optionally run 100% offline with data never leaving your device. This update also includes a ton of other features that bring deeper integration of your data into the AI portion, new ML tools, and various feature improvements across the board. I am happy to report that version 2.0 will be launching to the public on July 30th. In the meantime, I am very much open to feedback from other developers. I know we are all busy and appreciate your time.