Find AI · AI Device Privacy Policy

Last updated: 2026-01-21

Important Notes

This document explains Find AI’s AI/algorithm-related features, input data, processing approach, and on-device storage practices, aiming to meet Apple’s requirements on transparency and user control.

Core principle: data used for device finding (Bluetooth signals, location, and UWB when applicable) is processed on-device. We do not upload this data to our servers for analytics or training, and we do not track users across apps or websites.

This document covers device-finding related data and processing only.

1. App Scope

Find AI is a close-range device finding app. It discovers nearby Bluetooth devices, estimates distance, and provides map/navigation assistance to help users get closer to a target device. On supported devices and when pairing conditions are met, UWB (NearbyInteraction) may be used to provide more precise distance/direction hints.

2. AI / Algorithmic Features

Capability	Purpose	Generates Content	Cloud Processing
Distance estimation (RSSI → approximate meters)	Converts Bluetooth RSSI into an approximate distance and smooths short-term jitter	No	No (on-device)
Device type recognition (heuristic categorization)	Infers a device category (earbuds/watch/laptop, etc.) based on name and advertisement signals	No	No (on-device)
Navigation locating (UWB or Bluetooth fallback)	Outputs distance/direction and derived coordinates for map display and guidance	No	No (on-device)

3. Data Inputs Used by AI / Algorithms

3.1 Bluetooth

Peripheral identifier: system-provided peripheral identifier (CBPeripheral.identifier) used for on-device deduplication, pinning, and “last seen” records.
Advertisement data: e.g., LocalName, connectable flag, Manufacturer Data (for company ID parsing to assist type inference).
RSSI (dBm): used for distance estimation and signal strength UI.

3.2 Location & Sensors (for map/navigation and offline clues)

Current location: used to display the user on the map and to estimate distance to the last seen location.
Heading: used for bearing/direction estimation.

3.3 UWB (only when supported and the user enters the related navigation flow)

NearbyInteraction outputs: distance and direction vector (e.g., NINearbyObject.distance / NINearbyObject.direction).
Discovery token: NIDiscoveryToken to establish a NearbyInteraction session.

4. Processing (On-Device vs. Server)

All estimations and recognition are performed locally on the device.
We do not upload device-finding data (Bluetooth scan results, RSSI, location, or UWB data) to our servers for analytics or training.
The app does not use third-party AI/LLM APIs to process the above data.

5. On-Device Storage & Retention (AI Device related)

5.1 Pinned devices

Device identifier, name snapshot, inferred device type, pinned timestamp.
Last seen timestamp.
Last seen latitude/longitude (last_lat / last_lng) for offline “last seen location/distance” clues.

5.2 Device type recognition data

User-confirmed type overrides to improve future UI categorization.
Recognition feedback samples (predicted vs corrected + signals summary) to improve recognition in future; currently stored on-device only.

Retention: stored in the app’s local database (app sandbox) until the user unpins devices, clears app data, or uninstalls the app.

6. Third Parties & Sharing

We do not share device-finding data (Bluetooth signals, location, UWB data) with third parties for advertising or cross-service tracking.

If subscriptions/in-app purchases are enabled, transactions are handled by Apple. We may read purchase status on-device to unlock features.

7. User Controls & Transparency

Permissions: users can disable Location/Bluetooth permissions in iOS Settings; related features will be limited or unavailable.
Accuracy notice: distance/direction are estimates and can be affected by obstacles, reflections, interference, and device hardware differences.
Intended use: only for locating devices owned by the user or with authorization.