Hooshang

We did not want a wrapper around GPT with a Persian keyboard, something built for how people there actually think, talk, and ask.

Hooshang became the #2 native chatbot in the Persian market. Over a million installs in under six months. We brought top models into chat, image generation, and AI agents (what we called Modes) and made them simple enough for someone who's never heard the word "prompt."

We gave Hooshang a face. Duolingo figured out something the rest of the industry ignored for years: people form habits with characters, not interfaces. Duo the owl is annoying on purpose, and it works.
We took that instinct somewhere warmer. Hooshang is a simple, friendly presence that gives the app emotional weight without the guilt trip. People don't open a tool. They come back to someone familiar.

People don't open a tool. They come back to someone familiar. The UX writing carried that forward: warm, a little playful, high stickiness.

And before most apps were even thinking about it, we brought domain-specific AI skills into Hooshang, trained on specialized datasets, built for the moments when a generic answer just doesn't cut it.

General AI is flexible. It's also vague. And for anything sensitive. health questions, legal concerns, financial decisions, vague doesn't build trust.

ChatGPT solved part of this with GPTs. Powerful idea. But there's a catch: you have to know they exist, remember to open them, and figure out which one applies to your situation right now. That's a lot of cognitive load to put on someone mid-conversation.

We made a different call. Modes come to the user, not the other way around. When the conversation calls for a specialist, Hooshang switches into one. No navigation, no hunting, no deciding. The right expert shows up exactly when it's needed.

Flexibility traded for precision. Generic traded for trustworthy. There when you need them, without you ever having to ask!