Uzbek deserved a better app. So we're building it.

Why now

Uzbekistan is opening up faster than its language tools. Record foreign investment, more than 17,000 foreign companies, thirty-plus foreign university campuses, a 54% jump in international students in 2025, and an IT Park with 2,800 resident companies have brought hundreds of thousands of people to a country whose state language most of them cannot read yet. Since 2023 official life runs in Latin-script Uzbek. The old assumption that Russian is enough is fading in offices, clinics and classrooms.

The apps that exist treat Uzbek as one more row in a spreadsheet: the same forty units, translated, with English explanations and tourist vocabulary. That is not what a Chinese engineer on a petrochemical site, an Indian medical student on rotation, or a Russian-speaking developer registering a company needs.

What AI-native means here

Uzbekcha has no fixed course. When you open a lesson it is generated for you from three things you told us on day one: your native language, why you are learning, and your level. That changes everything downstream.

• Compared with your language. At launch the lessons are in English, but every comparison is written for your first language. A Turkish speaker (who already knows how suffixes work), or a Chinese speaker (who is meeting agglutination for the first time), or a Russian speaker (who knows Central Asia but not Turkic grammar) each gets a different explanation. Next, whole lessons in Russian, Chinese, Turkish, Hindi and Korean.
• Vocabulary from your life. Work, study, family, or the road, and inside "work", your industry.
• Grammar shown, not tabled. Every Uzbek word is split into root and suffixes so the pattern becomes visible: kitoblarimizda = kitob + lar + imiz + da.
• Audio everywhere. Every word, sentence and dialogue is spoken. Slow, normal, or native speed.
• A conversation partner. Suhbat lets you role-play real situations with the AI at your level and get corrected without embarrassment.

Large language models are famously shaky at Uzbek morphology on their own. Uzbekcha does not ship raw model output: generated Uzbek passes through a morphological validator that checks suffix order, sound changes at suffix boundaries and Latin-script consistency, and is grounded in verified grammar and dictionary sources rather than the model's memory.

Latin script, and only Latin script

Uzbek is written in a 29-letter Latin alphabet. It has been the official script since the 1990s and mandatory in state institutions since 2023, and the presidential administration switched fully in 2025. Uzbekcha teaches the Latin script exclusively, and shows the Cyrillic form only where you are likely to meet it, on older signs and documents.

Who we are

Languages at launch

English at launch: the app and every lesson are in English, and every grammar point is compared with your first language. Russian, Chinese (Simplified), Turkish, Hindi and Korean come next, then Arabic, German and French.

For universities, employers and institutions

If you onboard foreign staff or students who need Uzbek, we would like to talk. Cohort licences, a supervisor dashboard, and vocabulary tuned to your organisation are on the roadmap. Write to [email protected].