The AI Interviewer tab is hidden for studies using the Survey engine, since surveys don’t have conversational questions.
Interviewer mode
Find it under Advanced settings on the AI Interviewer tab.Personality & guidelines
- Description (up to 900 characters) — the interviewer’s persona: tone, role, and style. For example, a warm UX researcher, or a neutral market-research moderator.
- Guidelines (up to 3,000 characters) — behavioral rules for the whole interview. When set, your guidelines replace Versive’s defaults, so include everything you care about (probing style, what to avoid, how to handle off-topic answers).
Study context
Add Study context (up to 3,000 characters) to give the interviewer background about your study, product, audience, or terminology. For example: “We’re studying how small business owners use our invoicing tool. Most participants manage billing themselves and send fewer than 20 invoices per month.” This optional background helps the interviewer ask more relevant questions and follow-ups in both traditional and realtime voice interviews, including exploratory questions. It supplements your Description and Guidelines. Leave it blank or clear it to run interviews without additional study context. Study context is loaded by the interviewer on the server and excluded from the public study data sent to participants’ browsers. The 3,000-character limit also applies to API and assistant updates.Memory
Turn on Memory to let the interviewer learn from completed interviews in the same study and ask sharper, more informed follow-up questions as responses accumulate. After each completed live interview, Versive extracts small observations — emerging findings, follow-up questions that worked, threads participants left open — and folds them into a per-study briefing. In later interviews, the interviewer uses that briefing to pursue unresolved threads, probe findings that need confirmation, reuse probes that landed, and skip ground that’s already well covered. Good to know:- Off by default; toggle it in Settings → Memory. The same tab shows what the interviewer has learned so far, grouped by question, to members who can see the study’s results.
- What should the memory focus on? (up to 300 characters) — optional. Steers what gets remembered from each completed interview. It applies to interviews completed after you save, for example “Capture price sensitivity and the reasons behind it; ignore feedback about the app.”
- Any rules for using it in interviews? (up to 300 characters) — optional. Added to the interviewer’s memory briefing. It applies to live interviews as soon as you save, for example “Don’t bring up competitors from memory unless the participant mentions them first.”
- Only completed live interviews contribute — simulated interviews don’t feed memory.
- Findings are treated carefully: a pattern needs multiple independent confirmations before the interviewer treats it as established, and when participants disagree, the interviewer probes to resolve the disagreement rather than assuming.
- The interviewer never reveals or attributes what earlier participants said — memory shapes which questions get asked, not what gets shared.
Question rephrasing
By default the AI may rephrase the next question so it flows naturally from the conversation (“You mentioned onboarding was confusing — walk me through what happened when…”). You can:- Turn rephrasing off entirely if exact wording matters (e.g. standardized instruments).
- Add rephrasing instructions (up to 300 characters) to constrain how questions may be reworded.