What it does
- Answers questions using only the documents you make available: regulations, forms, decisions, guides, fee schedules, catalogues, manuals.
- Cites the source under every answer: the name of the document and the page, so the visitor can check it.
- Says when the answer is not in the documents, instead of improvising, and points the visitor to the institution.
- Replies in the language of the question, Romanian or English, to any number of visitors at once.
Who it is for
Public institutions
Town halls, county councils, agencies, universities, hospitals: the same questions arrive every day, and the answers are already in the published documents. The assistant takes the questions; the office keeps the decisions.
Companies with a lot of documentation
Manufacturers, distributors and service providers with catalogues, technical sheets, warranty terms and manuals. Customers get the right page of the right document, without waiting for an email.
Anyone who answers the same questions every day
Schools, associations, clinics, housing administrators. If the answers exist on paper, the assistant can give them; if they do not, it says so.
Installed in three steps
- 1
You upload the documents
The people you designate put the documents into a dedicated workspace: PDF, Word, Excel, XML, text, scanned images. Add a document and the assistant can answer from it; remove it and it no longer can.
- 2
You add one line to your site
You receive a code snippet configured with your workspace, language, colours, position and greeting, and add it to the pages where the assistant should appear.
- 3
You display the note for visitors
A short text next to the assistant, provided by us, says that visitors are talking to an automated assistant, that answers come from published documents and where the privacy policy is. Then the assistant goes live.
Once you accept the offer, we set it up with your team and adjust the tone and the note together.
What it never does
- It does not browse the internet and uses no source other than your documents.
- It takes no decisions on your behalf and issues no documents, opinions or approvals.
- It files no requests or petitions and has no access to your internal systems.
- It asks visitors for no account, name or other identifying data, and stores none.
- It is not used to train AI models.
Where the data is
- Hosting on Amazon Web Services in Frankfurt; the AI models run through Amazon Bedrock in EU regions. No transfer outside the EU in normal operation.
- Your institution has its own isolated workspace, with its own search index and storage.
- Visitor conversations are kept for 30 days, for technical support and for your own review, then deleted; earlier on request.
- No tracking scripts: the snippet keeps the current conversation in the visitor’s browser and sets no cookies.
With the look and tone of your site
Colours, position, greeting and the note shown to visitors are configured for you. The tone of the answers follows the instructions you give the assistant: formal for an institution, plain for a shop. Changes are made together with us, without touching your site again.