After asking a question, Hurai searches its own private legal database. This database contains Dutch tenancy law (Title 4, Book 7 of the Civil Code) and some other relevant laws and decisions. It then analyzes the uploaded documents to see whether they contain relevant provisions. Finally, it bases its answer on all information provided.
Hurai uses a protected version of ChatGPT offered by Microsoft, running on private Microsoft servers in the European Union. As a result, OpenAI does not see any of the data sent. All data traffic is encrypted and no call or document data is stored for training purposes. Hurai aims to train its own language model and move away from external providers.
Hurai does not store any call or document data and this data remains on your computer as long as the session is active. When you close your tab, all information is deleted. We do offer the option to download your conversation in a PDF file.
No, Hurai does not use your data to train models. This is also not possible, as we do not store your data.
Our database consists exclusively of legal sources from Dutch tenancy law and the system is limited to this information when answering questions - in addition to any uploaded documents. This keeps the chance of hallucinating to a minimum. Our system can properly determine whether it has relevant information in the database or not. During testing we have not yet encountered that Hurai is hallucinating, but we emphasize that you are and remain responsible for your own rights and obligations.
