FAQ
Frequently asked questions
Everything you might want to know before asking.
Artificial intelligence
- Does the AI do the analysis?
- No. The AI SUGGESTS codes, themes and relations; accepting, editing or rejecting them is yours. An unaccepted suggestion never enters a matrix, a count or a report. This is not a UI choice but the data model itself: pending suggestions live in a separate table and are not codings until accepted.
- Can the AI fabricate quotes?
- Three separate layers prevent it. First, when the model returns a quote we verify the text really occurs in that segment; if not, the suggestion is dropped. Second, the position of the excerpt is computed on the server — we never ask the model for character offsets, because models cannot count characters. An unanchored suggestion CANNOT be accepted and its count is reported to you. Third, in reports the model does not write quotes at all: it cites numbered excerpt ids and the text is resolved from the database.
- Can I turn the AI off?
- Yes, in the study settings. With it off, no text from that study reaches the analysis service and the product still works fully — manual coding, themes, matrices and reports all remain available. If your ethics board restricts AI use, this is a checkbox.
- Do I need to declare AI use in my paper?
- Most journals now ask for it. The panel records the origin of every coding: done manually, accepted as suggested, or accepted after editing. Codebook operations are kept with their rationale too, so you can write your methods section from that record.
Data and privacy
- Where do my interview transcripts go?
- Uploaded files and extracted text are stored in your own account space. If AI analysis is enabled, the analysed segments are sent to the model provider; if it is disabled, no text leaves at all. The analysis service does not retain what it receives.
- How do I protect participant identities?
- Each participant has a code (P1, P07) and an optional pseudonym. Matrices, reports and exports use the code. We recommend pseudonymising names in the transcript before uploading.
- Can I export my data?
- Yes. The codebook, quote table and full report download as Word and Excel. Every row in the quote table carries its excerpt id, so the table can be traced back to its source.
Files and formats
- Which file formats can I upload?
- Word (.docx), PDF, plain text (.txt, .md), subtitles (.vtt, .srt), CSV and Excel. The legacy .doc format is not supported; saving as .docx in Word is enough.
- Can I upload scanned PDFs?
- No. We do not OCR scanned PDFs, and we say so plainly: a half-working OCR introduces silent errors into the transcript, and in qualitative analysis a single wrong word can change a finding. If you upload one, the panel detects it and tells you explicitly.
- Can I upload audio or video for transcription?
- Not today; the panel works with text. You can upload the output of your transcription tool (a Word or subtitle file) directly.
Plans and billing
- What is the unit — users or documents?
- Documents analysed with AI per month. Manual coding is unlimited on every plan. Per-seat pricing would be wrong for qualitative work: one researcher may code 40 interviews for a thesis while an eight-person team processes 20 documents a month.
- Does re-running a document consume my quota twice?
- No. Quota is unique per (account, period, document); and if the document has not changed, the analysis is not re-run at all — the previous result is reused.
- Can I run a real study on the free plan?
- For a pilot, yes: one study, ten documents and unlimited manual coding. At thesis scale you will want the Academic plan.