What is a cited answer?
Along with the answer, the system returns the documents and pages it used. The user opens the source, sees where the sentence appears in the document and can reject the answer if needed. That is where verifiability comes from: you share not just the claim but what it rests on.
Respecting permissions
The second requirement for a business assistant is that it never answers from a document the user isn’t allowed to see. The permission check has to happen before the answer is generated; otherwise the system can leak the content of a restricted document in the form of a summary.
Where it really pays off
The uses that deliver most in practice:
- Comparing the same clause across many contracts
- Summarising a long report or specification
- Pulling together the history of a topic across files
- Producing structured output from documents (tables, lists, draft reports)
Where does the data go?
Three questions to ask before using AI on business documents: where is the data processed, is it used to train models, and how are organisations kept apart from one another? In mindfile, documents are processed only to answer your questions, are not used to train general models, and each organisation’s data is isolated in its own space. AI providers work under a Zero Data Retention policy, and the service runs on servers in Türkiye.
FREQUENTLY ASKED QUESTIONS
What if the AI gives a wrong answer?
That is exactly what citations are for: you can open the document the answer rests on and check it. Not producing claims without a source is a design rule; where there is doubt, the decision stays with the person reading the document.
Does it work on scanned documents?
Yes, after OCR. Once image-based documents are converted to text, both the search and the analysis layers work on that content.
