Every caller, answered like they reached the business's best employee.
Umay is an AI assistant set up per business: it speaks natural Turkish with callers, answers only from that business's own knowledge, and completes the work during the conversation — booking appointments, opening records, checking orders through APIs into the business's systems, defined at setup. It follows the business's rules, and hands the call to a human when it should.
On web and desktop today; the phone channel is on the roadmap. The demo speaks Turkish — that is the product. Questions in English? Get in touch →
Umay is built by Luvita Teknoloji, a software product company in Türkiye.
How it works
Umay is not a voice skin over a chatbot — it is an end-to-end conversation assistant: it listens, understands, and does the work.
Listens and understands
- Fluent Turkish speech recognition
- Stops and listens when the caller interrupts
- Discloses that it is an AI in its very first sentence
Does the work
- Connects to the business's systems over APIs — integrations defined at setup
- Runs the business's logic mid-conversation: checks availability, books appointments, opens records
- Before confirming, reads names back, phone numbers digit by digit, plates letter by letter
Hands over to a human when it should
- Never invents an answer outside the knowledge base
- Transfers angry callers and sensitive topics to the business's staff
Why Turkish-first
Generic multilingual bots are measurably weaker in Turkish — an agglutinative language where phone numbers, dates, plates and amounts must be spoken the Turkish way. Every layer of Umay is tuned for Turkish business conversations and validated by measurement, not by the model's own claims.
Natural Turkish speech
- A normalisation layer built specifically for Turkish
- Phone numbers, dates, times, plates and amounts read out correctly
- Backed by tests, not by the model
Speaks from the business's knowledge
- Answers come only from the business's own documents
- Outside them it doesn't guess — it hands over to a human
Integrated with business systems
- Appointments are just one example
- Business logic runs through API calls defined at setup
- CRM records, order lookups: any action the business defines
Read-back assurance
- Names, phone numbers, plates, days and times
- Read back and confirmed before anything is saved
Reliable human handover
- The assistant knows when to stop
- Defined safety triggers transfer the call to staff
Measured quality and cost
- Latency and cost logged for every conversation
- Reported in appointments booked, not characters generated
About
Umay is a Turkish-first AI call assistant developed by Luvita Teknoloji Enerji Yazılım San. ve Tic. Ltd. Şti. (luvita.tr).
Luvita's founder, Mert Ertuğrul, is a software engineer with close to fourteen years of experience in banking and finance. He spent seven years at Türk Eximbank — first as a financial analyst in the Bank Analysis unit, then in the Financial Institutions unit managing banking software projects as a product manager and developing software on those projects himself. He then spent five years at Channel Capital Advisors, a London-based trade-finance firm, leading the transformation of its enterprise systems to a microservice architecture; most recently he was a senior software engineer at ING Hubs Türkiye — ING's global technology centre — on the team building banking solutions for ING Italy, and left to found Luvita. He holds an MSc in Information Technologies from Sabancı University and is an Oracle-certified Java expert.
Hear it for yourself
In the demo, Umay answers as the customer line of a fictional bank: report a lost card, ask for a branch appointment, ask about rates, interrupt it mid-sentence — and watch how it responds, live. The conversation is in Turkish; that's the product. Banking is just one example — the same platform is programmed to a business's own workflows.