AI replies to Uzbek voice messages and document photos work right in the chat: the AI inbox transcribes a voice message in Uzbek and reads a passport, a receipt or a waybill, with no «could you send it as text, please». In Uzbekistan this matters: customers routinely record questions by voice and send document photos instead of typing. Messages from Instagram, Telegram and WhatsApp arrive in one inbox, where the AI employee (AI agent) grasps what the enquiry is about, mixed Uzbek-Russian speech included, replies in the customer's language and saves the recognised data in the conversation card. If a document cannot be read or the question is complex, the conversation goes to an operator with the attachment and the context, so the customer never sends the file twice. A voice or photo enquiry is handled as fast as a text message, at any hour, with no separate person sorting attachments.
Why this matters in Uzbekistan in particular
Customers in Uzbekistan communicate by voice and photo on a massive scale: they record a voice message in Uzbek instead of a long text and send a photo of a passport, a receipt or a waybill. Most global inboxes and aggregators do not understand this: the voice message goes unheard, and the photo has to be sorted by hand.
| What the customer sends | Typical case | What the AI does |
|---|---|---|
| A voice message in Uzbek | Saying it is easier for the customer than typing it | Transcribes the speech into text and answers to the point |
| A document photo | A passport for paperwork, a receipt for a refund, a waybill for reconciliation | Reads the data from the photo and saves it in the conversation card |
| Mixed speech | A voice message in Uzbek with Russian words | Grasps the meaning and replies in the customer's language |
| A voice message in Russian | Some customers record their questions in Russian | Transcribes it and replies in Russian |
How it works
- The customer sends a voice message or a photo in Instagram, Telegram or WhatsApp.
- The AI transcribes the Uzbek speech into text and reads the data from the photo.
- It works out what the enquiry is about and answers to the point, in the customer's language.
- It saves the data in the conversation card and passes complex cases to an operator with context.
All of this runs in the single inbox. How voice works in the separate call channel is described on the voice AI page.
What global solutions lack
Global helpdesk platforms and messenger aggregators are built for text in major languages. Uzbek voice and document photo recognition right inside the chat are a rare local edge. Zukko.AI is a platform of AI agents for business with its own Uzbek voice, which global models do not provide.
How to launch
- Connect Instagram, Telegram and WhatsApp to the single inbox: voice messages and photos from all three channels will flow into one feed.
- Upload the knowledge base: price list, delivery and return terms, answers to frequent questions. The AI uses it to answer the transcribed voice message.
- Set the escalation rules: which enquiries and documents go straight to an operator and which the AI closes itself.
Onboarding takes from 2 hours for small business and up to 30 days for enterprise with integrations. In the first week the operator reviews the enquiries handed over to them and adds to the knowledge base, so the AI closes more questions on its own.
What it gives the business
- You keep the customers who send voice messages: the AI hears and understands them.
- Document photos are read automatically: no manual sorting.
- Uzbek voice is not a problem: it is transcribed like text.
- The data lands in the conversation at once: the operator sees what was recognised.
The full picture of the single inbox is in the omnichannel guide.
Frequently asked questions
Does the AI understand voice messages in Uzbek
Yes. It transcribes Uzbek speech into text and answers by meaning, mixed Uzbek-Russian speech included.
Can the AI read a document photo
Yes. It reads the data from a photo of a passport, a receipt or a waybill right in the chat.
Do I need to ask the customer to type instead
No. The AI handles both voice and photos, so the customer has nothing to retype.
Does it work in Instagram and Telegram
Yes, in Instagram, Telegram and WhatsApp, in one inbox.
What happens if the AI cannot read a document
The conversation goes to an operator with the context and the attachment, so the customer does not send the file again.