Team notes and product updates
Here we share how the model is built, how companies and regular users work with it, what we think about privacy, and why we build a regional product in particular.
Latest materials
All posts
EYWA public launch: an AI assistant with local infrastructure
We opened EYWA as a separate consumer product. We tell why we brought the assistant out of the corporate perimeter and how it differs from familiar chats.
Why we have our own GPU cluster in Uzbekistan
The story of moving from external APIs to our own inference stack and what that changed for our clients in terms of latency and privacy.
Eywa One in simple words
Briefly on how CRM, warehouse, finance, contact center and chat assistant merge into one platform without a zoo of services.
Artifacts and projects: how we rethought continuing work
Project memory, artifact versions and workspaces. We tell why these three together change the feeling of working with an assistant.
How we built privacy policy in practice
The technical decisions behind 'your data stays within the local perimeter'. No marketing — just architectural details.
Russian, Uzbek, English: how we trained regional context
What exactly we improved in the model for local scenarios and why standard frontier models don't always work well with regional specifics.
Detailed materials are published gradually. If you want notifications about new articles and updates — write to us at hello@eywa.uz.
Practice
Open the product, don't read about it
The blog is a nice addition, but the best way to understand EYWA is to open a chat and ask your first real question.