AI Generator Studio
A creation studio that brings fifteen separate AI tools together in a single app.
- iOS
- Android
- 2025
- My own product
Screenshots
Click an image to enlarge it.
About the app
When AI tools were just becoming widespread, most apps did exactly one thing: either text to image, or text to video. I asked a different question — why shouldn't all of them live in a single app? AI Generator Studio came out of that idea and brought 15 separate AI features together in one interface.
Technically, the real work wasn't writing the models but unifying different providers behind a single interface. Every feature used a different service, a different request format and a different response time. On the app side I reduced all of them to a shared "generation job" abstraction: the user starts a request, the job enters a queue, and a notification arrives once the result is ready.
On top of that I built a credit system. Because each feature had a different cost, a credit-based model made more sense than a flat subscription; credits are bought through in-app purchase, reserved before generation starts, and refunded if the job fails. That way users never lost credits to third-party service errors.
A 2025 project, written before AI-assisted coding tools became widespread. AI was the subject of the product, not the development tool.
The app isn't live today: I ran out of time to keep it going on my own. Fifteen separate AI features meant fifteen separate providers; each one's interface, pricing and error behaviour kept changing, and keeping up with those changes was a constant job in itself. The code side worked fine — the problem was maintenance capacity. The lesson I took from this is that in a product with heavy external dependencies, the real cost isn't writing it but keeping it current.
Key features
- Text to image: prompt-based image generation with style options
- Text to video: turning short stories into video clips
- Image to video: animating a still image
- Text to speech: multilingual voice-over with previewable voice selection
- Voice cloning: a personal voice generated from an uploaded audio sample
- Podcast generation: a narrated podcast episode from text
- Avatar generation: an AI avatar personalised to the user
- Short film generation — from script to scene sequence
- AI Dating: chat with an AI character shaped around your interests
- AI Joke Machine and other entertainment-focused generation tools
- Credit system: per-feature cost, reservation and automatic refund on failed jobs
- Asynchronous generation queue: long-running jobs in the background, results delivered by notification
- Generation gallery: past outputs stored and re-downloadable
Technologies used
109 technologies · 12 categoriesLanguages and Platform
- Swift (iOS)
- Kotlin (Android)
- Java (Android)
- iOS SDK
- Android SDK
- Native development on both platforms
UI and Application Layer
- UIKit / SwiftUI (iOS)
- XML Layout / Jetpack Compose (Android)
- UICollectionView (gallery)
- RecyclerView (gallery)
- Auto Layout / responsive layout
- Audio preview player
- Loading and progress states
Backend and Data
- Firebase Firestore
- Firestore Security Rules
- Firestore transactions (atomic credits)
- Firebase Storage
- Cloud Functions
- Firebase Authentication
- Server-side business logic
- Real-time listeners
- Firestore offline cache
- Batch write
- Cursor-based pagination
- Denormalisation
- Counter fields
- Cloud Functions triggers
- Scheduled jobs
- Data migration
AI Providers
- Generative image/video APIs
- Text-to-speech (TTS)
- Voice cloning API
- REST / JSON integration
- URLSession (iOS)
- API key hiding (Cloud Functions)
- Prompt design
- ElevenLabs API
Real-time and Media
- Firebase Cloud Messaging
- APNs
- Notification permissions
- Asynchronous generation queue
- AVFoundation (iOS)
- Android media playback
- Image/audio picker (PHPicker / Photo Picker)
- Saving to the gallery (Photos / MediaStore)
Payments and Revenue
- StoreKit (iOS)
- Google Play Billing (Android)
- Consumable products
- Server-side receipt validation
- Credit pack configuration
- Reservation and automatic refund
Architecture and Patterns
- MVVM / MVC
- Swift Concurrency (async/await)
- Kotlin Coroutines
- GCD / background threads
- Job state machine
- Error handling and retries
- Idempotent credit accounting
- Modular feature architecture
Networking and Data Transfer
- HTTP client (Android)
- JSON parsing (Codable)
- Retry and timeout handling
- Offline state management
- Image caching
- Download and upload queue
- Multipart file upload
- Bandwidth-aware quality
Security and Privacy
- Keychain (iOS)
- EncryptedSharedPreferences (Android)
- HTTPS and ATS
- Auth token management
- Session refresh
- Permission rationale copy
- Account deletion flow
- Server-side authorisation checks
- Sensitive data masking
UI States and Accessibility
- Dark theme support
- Dynamic Type
- Safe area
- Empty / loading / error states
- Skeleton loading
- Pull to refresh
- Keyboard and focus management
- Deep linking (Universal Links)
- VoiceOver / TalkBack labels
Testing, Debugging and Profiling
- Xcode Instruments
- Android Profiler
- LLDB
- Logcat
- Memory leak tracking
- Network traffic inspection
- Reading crash reports
- Testing on real devices
- Testing across screen sizes
Tools and Release Process
- Xcode
- Android Studio
- App Store Connect
- Google Play Console
- CocoaPods
- Gradle
- Git
- TestFlight
- Play Console internal testing
- Version and build numbering
- Signing and provisioning profiles
- App Store review process
- App Privacy / Data Safety form
- Store assets and ASO copy
- App size optimisation
Idea, design, UI, code, app store release and post-launch maintenance — I ran the entire process end to end on my own. No team, no agency, no outsourcing.
Development notes
- Languages and approachThe iOS side was written in Swift and the Android side in Java and Kotlin. No cross-platform framework was used; both apps were built with their own platform's native tooling. I have been doing mobile development since 2020.
- No AI assistanceEvery line of code was written by hand. This project predates the spread of AI coding tools like Claude Code and Codex — nothing of the sort existed at the time.
- DependenciesA large number of open-source GitHub libraries and CocoaPods packages were used. Dependency management was handled with CocoaPods on iOS and Gradle on Android.
- Provider abstractionThe fifteen features were tied to different services; all of them were reduced to a shared “generation job” abstraction. API keys never reached the client — every call went through Cloud Functions.
- Testing and debuggingTested on real devices, across different screen sizes and under poor network conditions. After launch I read the crash reports and shipped fixes release by release.
- Release processI handled the review process for both stores on my own: App Store Connect and Google Play Console setup, privacy declarations (App Privacy / Data Safety), release notes, store assets, and fixes based on rejection reasons.
- Product ownershipIdea, design, development, backend, app store release and post-launch maintenance — I ran the entire process on my own.
Let's talk about AI Generator Studio
I'd be glad to walk you through the technical details, the problems I ran into and how I solved them.