ORA — AI Assistant & Creator

Overview

Started as a small photo app, now a full generative suite: chat, image, video, music, voice and writing tools. One interface, routed across DeepSeek, LLaMA, Kimi, Qwen and GPT.

0K+

users

0+

AI models

The problem

People were switching between a chatbot, an image generator, a video maker, a music tool and a transcription app, each with its own subscription. ORA puts them behind one interface and routes each request to the model best suited for it.

How it was built

Flutter on the client, a NestJS backend in Docker on AWS EC2 with OAuth2. The first release ran custom PyTorch pipelines, GANs and Stable Diffusion for photo editing. Version 2 added a routing layer over hosted models, DeepSeek R1, LLaMA 4, Kimi K2, Qwen and OpenAI GPT among them, so new models can be added without changing the app.

What it does

• 01AI chat across 20+ premium models in one conversation surface
• 02Image generation and editing: object removal, background editing, upscaling
• 03Video creation, music generation, voice and voice cloning
• 04OCR and transcription, writing and coding assistants, everyday AI utilities

Outcome

20,000+ users and 20+ models routed through one app after two years of evolving alongside the models themselves.