Frequently asked questions

What is ProSe?

ProSe is a legal AI assistant for Bulgaria. You ask a question in plain language and receive an understandable answer grounded in Bulgarian legislation and court practice. The name comes from the Latin legal term pro se — “for oneself”: everyone's right to represent and inform themselves.

Am I talking to a human or to artificial intelligence?

Answers are generated entirely automatically by an artificial-intelligence system — not by a human. This is also indicated in the chat itself. Treat them as an informational starting point, not as a lawyer’s opinion.

Are the answers legal advice?

No. Answers are informational and do not constitute legal advice. They are an excellent first step — understanding your situation, your rights, and possible actions — but consult a lawyer for important decisions.

What sources does the assistant use?

Current Bulgarian legislation and published court practice, including live search across court acts. The assistant relies on real sources, not general knowledge.

How much does it cost?

Registration is free. You choose a monthly or annual plan (annual is discounted) or buy top-up packs that never expire. Usage is measured in tokens — units roughly proportional to the length of questions and answers. See Pricing for an estimate of how many questions each plan includes.

Can I upload a document?

Yes — PDF, DOCX, or a photo (including scans, which are read automatically). The assistant explains the content, the clauses, and potential risks.

What is a company account?

A company account gathers a team around a shared token pool, with individual monthly limits and roles. It is available on plans that include the company-accounts option.

What happens to my conversations and data?

They are stored encrypted on servers in the EU and are not used to train AI models. You can export your data (GDPR) or delete your account from your profile page at any time.

How do you protect sensitive data like national ID or IBAN?

ProSe automatically detects and masks structured personal identifiers — national ID (ЕГН), foreigner number (ЛНЧ), IBAN, payment-card numbers, phone and email — in both your question and any attached documents before the text is sent to the AI system, so those identifiers do not leave our server (for the production service, hosted in Bulgaria). Additionally, person names and addresses are pseudonymized: replaced with codes (e.g. "ЛИЦЕ-1"), the AI system only works with the codes, and in the answer you see the real names are restored locally on our server. Your original text stays encrypted in your account and is visible only to you. Note: the free-text substance of your question (the description of your case) is processed by the AI provider so it can answer, and automatic detection may in rare cases miss a value (for example an unusual name, or a number split by spaces in a scanned document) — so please continue not to enter unnecessary personal data.

How do I know an answer was generated by AI?

We label this in several places: the chat carries a persistent notice that you are talking to artificial intelligence; every exported PDF bears a visible "AI-generated" line on each page plus machine-readable marking in the file's metadata. The labelling follows the transparency requirements of the EU Artificial Intelligence Act (Art. 50).

How do I cancel a subscription?

From “My account” → “Cancel plan”. The plan stays active until the end of the paid period, and pack tokens never expire.

I have another question — where do I write?

Through the contact form or, if you have an account, from the “Support” section, where you can also track your inquiries.