Challenge
Turkish users who wanted access to the best available AI models had to juggle several separate apps and subscriptions, and every one of them, ChatGPT, Claude, Gemini, Grok, was built English-first. None were genuinely fluent in Turkish language or cultural context, which meant constant app-switching and a worse answer at the end of it.
Approach
Rather than wrap one model, we connected several frontier models (OpenAI, Anthropic Claude, Google Gemini, and xAI Grok) behind a single native interface, built in FlutterFlow for a fast cross-platform ship, with Turkish language and cultural context as a first-class design requirement rather than a translation layer bolted on afterward.
What we built
- Multi-model access to OpenAI, Anthropic Claude, Google Gemini, and xAI Grok through their official APIs, presented as named, distinct assistants rather than one generic chatbot
- Turkish language and cultural optimisation across the interface and prompting, not just machine-translated output
- A single unified chat interface, so switching models does not mean switching apps or losing context
- Encrypted conversation storage
- Native iOS and Android builds submitted under the client’s own developer accounts
Outcome
Bilge AI is live on the App Store and Google Play, with more than 15,000 users accessing every major AI model in one app, in their own language, with no subscription-juggling or app-switching. The client owns the app, the developer accounts, and the API relationships outright.