Seyla

Inspiration

Talking to a camera is harder than writing. A full script helps with memory, but it also makes people sound like they are reading. When a take goes wrong, creators often restart the entire video, even though only one thought needed another attempt.

I built Seyla around a simpler idea: keep your place without putting a script between you and the camera.

What Seyla does

Seyla is a native iPhone anti-teleprompter for talking-head videos. You write a project as short beats, record several takes for each beat, keep the take that sounds most like you, and stitch the selected takes into one vertical video.

The core flow is:

Beats → record → keep → stitch → share

It is deliberately not an AI writer, captions app, or full video editor. The product is about helping you speak naturally while keeping the recording process structured.

How it was built

Seyla is built with Swift 6, SwiftUI, SwiftData, and AVFoundation for iPhone. Takes and exports are stored locally in the app’s Application Support directory. Raw video is never uploaded to the backend.

Supabase handles passwordless email authentication and account lifecycle events. RevenueCat powers the membership and Pro entitlement, with the RevenueCat customer linked to the authenticated Supabase user. The paywall supports purchasing and restoring a membership through the App Store.

The app is designed around service boundaries so camera capture, video assembly, Photos export, sharing, authentication, and monetization can be tested independently. This was important because a recording app has to behave reliably when permissions, storage, camera startup, or export operations fail.

Challenges and lessons

The hardest parts were making local media durable across relaunches, keeping older takes safe when recording another attempt, validating that every selected take was still readable before stitching, and making export failures recoverable without losing the stitched file.

The biggest lesson was that the product becomes clearer when the constraints are intentional. Seyla does less than a traditional editor, but every screen supports one job: helping the creator keep their place and sound like themselves.

Built With

• avfoundation
• ios
• revenuecat
• supabase
• swift
• swiftdata
• swiftui