AiPro – AI Chat Bot Assistant

AiPro is an all-in-one AI assistant for iOS and Android, built for DekMobile Dijital. It goes beyond a chat window: users can talk to a GPT-4-powered assistant, point the camera at an object or document, turn photos into editable text, generate images from a description, and translate speech and images in real time. The app is available in seven languages and runs on fully native clients.

The brief

The goal was a consumer AI app that could compete in a crowded category by doing more than text chat — and doing it fast. That meant native performance on both platforms, a backend able to orchestrate several AI services behind one conversation, and a subscription model that kept per-user AI cost under control.

What we built

Conversational assistant on GPT-4. Streaming chat with conversation history, prompt templates for common tasks (writing, email, research, summarising) and a model layer that allows switching models without an app update.

Image recognition and OCR. Users photograph an object, a sign, or a document; the app identifies it or extracts the text into an editable, translatable format.

Image generation. Text-to-image creation from the same conversation flow, with high-resolution export.

Real-time voice and image translation. Speech-to-text, translation, and text-to-speech chained in a single pipeline; camera translation for signs, menus, and documents.

Document assistant. Upload or scan documents and ask questions about them, summarise them, or extract structured data.

Multilingual UI. English, German, Portuguese, Italian, French, Spanish, and Turkish, with locale-aware prompts so responses match the user's language.

Subscription and usage control. Free tier with daily limits, premium tier with unlimited usage; per-user token accounting and rate limiting on the backend to keep AI costs predictable.

Architecture

• iOS: Swift and SwiftUI, native camera and document scanner integration
• Android: Kotlin with Jetpack Compose, CameraX for capture and scanning
• Backend: AdonisJS (Node.js) API orchestrating OpenAI chat, vision, image generation, speech, and translation services
• Web: Next.js admin panel for prompt management, usage analytics, and user administration
• Data: MongoDB for conversations, documents, and user data; Redis for session state, rate limiting, caching, and streaming queues
• Real-time: WebSocket-based response streaming for a live chat feel
• Monetisation: App Store and Google Play subscriptions with server-side validation

Technical highlights

• Single backend pipeline that routes a request to the right AI service (chat, vision, OCR, image, speech) based on input type
• Token and cost tracking per user and per feature, surfaced in the admin panel
• Content safety filtering on both input and output
• Response caching for repeated queries to reduce API spend
• Model-agnostic chat layer: providers and models configurable server-side

Outcome

AiPro shipped to the App Store and Google Play as a live, subscription-based product in seven languages. The modular backend has allowed new AI capabilities to be added — and underlying models to be swapped — without rebuilding the mobile apps.

Services provided

AI product architecture · LLM and multimodal AI integration · Native iOS development · Native Android development · AdonisJS backend development · Next.js admin panel · UI/UX design · Subscription implementation · App Store and Google Play launch