Many sports fans want to use AI tools like ChatGPT for sports betting insights, but there are concerns about the privacy and security of the sensitive prompts, especially when dealing with real-time data and proprietary strategies. Users also demand that their data and conversations remain fully confidential, without risk of interception or server-side leaks, and wish to access real-time sports odds and analytics on any device, even without internet connectivity.
This project created a privacy-preserving sports betting assistant by developing a compact, custom small language model that runs entirely on users' phones (iOS and Android), never sending prompts or data to external servers. The workflow pulls live sports odds via an API, converts that data from JSON to CSV to DataFrame, graphs it, and finally formats it as plain text readable by lightweight local AI. Leveraging ONNX and PyTorch, the model is optimized to operate entirely offline, giving users full control over their questions and predictions while ensuring prompt security.
.jpg)
.jpg)
The solution empowers users to interact securely and privately with an AI for sports betting, even on airplane mode or in settings where network security is a risk. By running the model locally, it prevents data leakage, giving confidence to high-stakes bettors and privacy-conscious users alike. The real-time data pipeline ensures up-to-date odds and analytics, while the project's open approach helps set a new standard for how AI and sensitive, high-value use cases like sports betting can be securely handled on personal devices.
