Frequently Asked Questions
Find answers to your questions here
Is our data used to train models?
No. Customer data is not used to train general (shared) models. Only with explicit written consent is it used for customer-specific fine-tuning, and those models are accessible only from your account.
Is on-premise installation possible?
Yes. On the Enterprise plan you can run our products in your own data centre or a private cloud in Türkiye. GPU needs vary by product; sizing is done in the discovery call.
How long does a pilot take and what does it cost?
A standard pilot takes four weeks and starts at 285,000 TRY. If the pre-agreed success metrics are not reached at the end of the pilot, part of the fee is refunded or extra time is provided.
How is the Turkish performance?
Our models are fine-tuned for Turkish and separately evaluated on agglutinative morphology, dialects and conversational language (abbreviations, typos). Turkish test results are shared in the model cards.
Which systems does it integrate with?
Ready connectors exist for CRM, ERP, contact centre, WhatsApp Business, e-mail, SSO, document management and data warehouse; other systems connect via REST API and webhooks.
How is the risk of wrong answers managed?
Generative answers are produced only from approved sources and cite them. If confidence is low, the matter is handed to a human. The error rate is also tracked with regular evaluation sets.
How does API pricing work?
Plans include a monthly ‘AI unit’ quota. One unit equals 1,000 tokens, one document page, one image inference or one minute of speech analysis. Overage is charged per unit.
Which sectors do you have references in?
We have 38 enterprise customers across banking and insurance, telecom, retail, automotive supply, logistics, healthcare and public institutions.
Could not find your answer?
Our expert team will be happy to help you.