Face Swap SDK for Telehealth Apps
Amigo AI SDK adds on-device face anonymization to iOS telehealth apps. Every frame is processed locally via CoreML and never leaves the patient's phone — so PHI exposure is eliminated from the video path, not just encrypted in transit. The per-frame API drops into any WebRTC stack (Twilio, Daily, LiveKit, Vonage, custom SFUs), runs at 30fps, and lets patients appear under a consistent synthetic identity for sensitive consults. $0.01 per session with 1,000 free every month.
Why on-device matters for telehealth privacy
Cloud face-swap APIs require uploading raw patient video — a textbook PHI disclosure. Even with TLS and a BAA, the server sees the patient's actual face, and every frame is a potential incident surface.
Amigo AI SDK reverses that model. Face detection, embedding, and swap all run on the device GPU. The WebRTC track you send upstream already contains the anonymized face. Your own video infrastructure, your vendor's SFU, and any observer in the network path only ever see the synthetic identity.
For group therapy, addiction medicine, gender-affirming care, and any consult where identity exposure is itself a harm, this is the difference between "encrypted PHI" and "no PHI to expose."
How do I add patient-privacy face swap to a telehealth app?
Add AmigoFaceSwapSDK via Swift Package Manager, initialize with your API key, enroll a synthetic target face to get a FaceLatent, then call processFrame(_:using:) on each camera frame before handing it to your WebRTC publisher.
import AmigoFaceSwapSDK
// One-time: enroll a synthetic patient identity
try await AmigoFaceSwap.initialize(apiKey: "your-key")
let anonLatent = try await AmigoFaceSwap.enrollFace(from: syntheticFace)
// In your WebRTC local video capture callback:
func onLocalFrame(_ pixelBuffer: CVPixelBuffer) {
if let safe = try? AmigoFaceSwap.processFrame(
pixelBuffer, using: anonLatent
) {
webRTC.publish(safe) // Upstream only sees the synthetic face
}
}Privacy guarantees at a glance
| Data | Leaves device? |
|---|---|
| Raw camera frames | No |
| Face embeddings (FaceLatent) | No |
| Biometric identifiers | No |
| Anonymous session count (billing) | Yes — no PHI |
Frequently asked questions
Ready to add face swap to your app?
1,000 free sessions every month. No credit card required. $0.01/session after that. Integrate in minutes, not days.