- Adaptive infrastructure for healthcare.
- Medical AI in hospitals.
- Voice AI in development.
Work — what we ship
About
"The best AI product is one the doctor doesn't notice. They just finish on time."
An applied AI lab in Tashkent. We build production AI for problems with real cost: clinical hours lost to paperwork, call centers in languages global vendors ignore, exploration and farming decisions still made on spreadsheets.
AvicAI started when Ulugbek watched doctors spend most of their day on documentation. The diagnosis took minutes. The paperwork took hours. That specific gap became the product.
Four products: one in hospitals, one on the phone, one under the ground, one in orbit.
Principles
What never changes while everything else does:
Automation has boundaries.
We don't believe AI should replace people. We believe it should replace repetitive Tuesday afternoons.
Good enough isn't.
Close enough works for weather forecasts. Not for healthcare.
If it's complicated, it's not finished.
Healthcare is already busy enough. We obsess over making every interaction feel obvious, so people spend less time learning software and more time helping patients.
Quiet confidence.
We don't need to call ourselves revolutionary. If the work is good, people usually notice.
Notes
Field notes
Working notes from the lab. What broke, what shipped, what we measured. Occasional. No hype.
SubscribeThe demo is the pitch.
Twenty minutes, a live system, your data if you bring it.