Quria Discovery AI FAQ
Here we’ve compiled a list of frequently asked questions about Quria Discovery AI. If you can’t find the answer to your question, or if you’d like to schedule a demo or try out Quria Discovery AI, please get in touch with us!
Can users’ clear questions and answers if the chat is used on a public library computer?
As a user, you can end the session at any time using the button at the top. This resets the chat to its initial state, and no previous conversation will be visible. If the user does not do this, the chat will automatically reset after a short period of inactivity.
Does the AI use external sources for information outside of Quria?
Only the connected sources of information used in Quria in Arena such as covers and opening hours.
Can the library populate the chat with information that is not currently in Quria, such as opening hours or other local library details?
Yes, this is possible. As part of the implementation project, we will ask a set of standard questions that libraries typically want the chat to handle, opening hours being one of them.
For example, recurring questions that staff frequently receive can be preloaded into the chat. This helps reduce repetitive inquiries and improves efficiency for the library team.
Does the chat work with static or dynamic data sources?
A combination of both. Static data sources include information provided by the customer in text form during the implementation project. Dynamic data include catalogue data that is harvested evey night, holding information that is fetched in real time when a title is chosen and opening hours that are fetched from the Opening hours service that is used in Arena.
How does Quria Discovery AI handle personal data and borrower privacy?
The service does not require users to log in and therefore does not need to identify who they are. The chat is not connected to borrower accounts, borrowing history, or personal profiles, and conversations are not stored.
If a user chooses to enter personal data, it may be processed temporarily in order to generate a response. However, users are clearly advised not to share such information in the chat. Conversations are retained pseudonymously for a short period and are then automatically deleted. The data is not used to train the AI model.
All data is handled within the EU, and the underlying AI model is contractually restricted from using conversations for its own learning or further training. This approach minimizes the processing of personal data and ensures that the solution is designed from the outset to protect user privacy.
If conversations are not stored, how is usage data and statistics generated?
The service is anonymous by design — no login, no user account, no profile of the visitor.
Conversations are retained pseudonymously for a short period and are then automatically deleted. Statistics — top topics, languages, peak hours, search success rates — are generated by aggregating these records. No individual conversation content persists in the analytics layer.
What data is used to generate recommendations for broad queries (e.g., books for a 9-yearold)?
Exclusively the library’s own catalogue. The agent translates the natural-language request into a structured search with filters for audience, age, language and genre, and matches it against the catalogue using a combination of keyword and multilingual semantic search. No external book databases at query time, no data from other libraries, and no profile of the asker.
What happens if a user enters personal information in the chat that should not be there because of privacy concerns?
Privacy by design here means four properties working together: no login, so a conversation cannot be linked to an identifiable person; no profile of the user; short retention with automatic deletion; and a clear advisory in the interface that sensitive personal data should not be entered in free-text.
What happens if a borrower asks something unrelated to the library’s services?
This does not create any issues. In such cases, the chat responds in a polite but firm manner, explaining that it cannot assist with that type of request and guiding the conversation back to library-related topics.
Which languages does the chat support?
The chat can be configured and adapted based on the library’s needs. In the current demo version, the following languages are enabled: German, English, Spanish, French, Somali, Finnish, Swedish, Turkish, Ukrainian, Urdu, Arabic, Pashto, and Persian.
Additional languages can be added.