AI applications
Social Copilot
An iOS communication assistant that combines a message, relationship context, and writing preferences to suggest three replies for the user to review and copy.
One contact and one message
Social Copilot is a working MVP for drafting replies. The user provides a message, chooses a contact and a communication goal, and receives three suggestions. The app does not access WeChat or SMS conversations or send replies automatically.
Messages can be typed, pasted, or dictated. Users can also attach a screenshot as visual input to the model. Contact records hold relationship details, notes, and memories maintained by the user. Writing preferences are visible and editable.
Implementation
The iOS client uses SwiftUI, Speech, and AVFoundation. Local data is stored in protected atomic JSON snapshots, with data migration and failure recovery.
Model requests pass through a Node.js proxy that keeps API keys on the server. The proxy applies request-size limits, timeouts, and rate limits without logging prompts or request bodies. Verification includes iOS unit tests, UI tests, and backend tests.
The implemented flow covers selecting a contact, entering a message, generating suggestions, copying a reply, and optionally saving the history.
SwiftUI · Multimodal AI · Speech · Node.js