AI-assisted authoring, conversational follow-up on open-ends, anonymous email distribution, wave-over-wave trending, and export that lands in SPSS as a finished dataset. This page is the full contents of the plan — not a summary, and not a list of what other plans are missing.
AI drafts the questionnaire, reviews it for bias, and flags weak questions before you field.
Answers arrive auto-themed and quoted in a readout, instead of as a column to read by hand.
Field to your own list as often as you like. Panel sample is bought separately, as needed.
The single biggest loss in survey research is the thin open-end you can't go back and probe. AI Interviewer asks in the moment, in context, so you get the reason behind the rating instead of “it's fine.”
Open-ends come back clustered into themes with counts, sentiment tagged, and quoted in a paste-ready executive summary. What used to be a week of tabbing is a scroll.
SPSS export carries variable names, value labels, and measurement levels. Run the syntax and the file opens ready to analyse — no re-declaring 60 variables by hand.
Link repeated runs into a series and every metric charts wave over wave. Send to a named list, track who has responded, and remind the rest — without ever tying a response back to a person.
Allowances are generous by design — they exist to stop runaway automation, not to meter ordinary use.
When a respondent gives a thin open-end, the AI asks a follow-up in the moment — the way a moderator would. Probes are generated against the question's own context, so a one-word answer becomes something you can actually code.
Describe what you need to learn and get a drafted questionnaire — question types chosen to fit, scales set, logic scaffolded.
Reads a finished draft for leading, double-barrelled, and biased items, plus broken scales and missing options — each flag with a one-click fix.
Auto-coded clusters across every open-end, so a thousand free-text answers arrive as themes with counts rather than a column to read by hand.
A paste-ready readout of the findings — what moved, what didn't, and what the open-ends say — written from the actual response set.
Paste a questionnaire you already run — a Word doc, a competitor's export, a list in an email — and get a working survey draft to edit.
The area of the platform that has moved most this year.
The response record genuinely carries no identifying field — no IP address, no user agent, no email. That is a property of the data model, not a setting somebody has to remember to switch on, and it is a stronger position than most tools in this category can state.
Minimum-N gating closes the other half: an author who filters down to a three-person slice sees a suppression panel, not three people's free-text comments. It guards against accidental disclosure through the results interface — it is not a technical guarantee against a determined author with developer tools.
| Feature | Allowance | Resets |
|---|---|---|
| AI Author | 200 | Monthly |
| AI Reviewer | 300 | Monthly |
| AI Themes | 50 | Monthly |
| AI Exec Summary | 50 | Monthly |
| AI Parse | 50 | Monthly |
| AI Interviewer probes | 500 | Per survey |
| Active surveys | Unlimited | — |
| Questions per survey | Unlimited | — |
| Responses from own audience | Unlimited | — |
| Collaborators per survey | Unlimited | — |
| Panel completes | Pay-as-you-go | — |
Monthly allowances reset on the first of the month. The AI Interviewer allowance is per survey, so every new study starts with a full 500.
Per survey. Every new study starts with a fresh 500 probed responses, so a big quarterly programme doesn't eat the allowance a smaller study needs the same month. Probes are only spent when the AI actually decides a follow-up is warranted — a well-answered open-end costs nothing.
A fixed follow-up fires for everyone, including the people who already answered well — which lengthens the survey and annoys your best respondents. AI Interviewer reads what was actually written and only probes when the answer is thin, then asks something specific to it rather than a generic “can you tell us more?”
Yes, and the mechanism matters. Each recipient gets a unique invite link. When they submit, the response is stored exactly as any anonymous response would be — the invite token is never written to it, in any form, including hashed. A separate write marks that person as having responded, timestamped only to the hour. So the system knows that somebody responded, which is all a reminder needs, and cannot reconstruct what anybody said. There is no column, relation, or index linking a recipient to a response.
Not through the interface. Set an anonymity floor on the survey, and any filter combination that narrows below it — department plus tenure plus location, say — suppresses the charts and open-text instead of rendering them. Worth being precise with your own people: this is a guardrail against accidental disclosure in the results view, not a technical guarantee against somebody determined with developer tools.
Out, in whatever shape you need. SPSS with a labelled CSV and syntax file, XLSX, or CSV — each in standard, condensed, expanded, or raw column shapes, and scoped to whatever filter you have applied. Signed webhooks push responses to Slack, Sheets, Zapier, Notion, or Airtable as they arrive. Your data is yours; there is no export gate and no vendor lock-in.
Ten completes is the practical minimum. Panel sample is pay-as-you-go on every plan, including Free, so there is no subscription gate on buying sample and no minimum commitment — buy as few or as many completes as the question warrants.
No. Collaboration is per survey — invite a colleague at view or edit access on any study, as many as you need, with no per-seat maths. If your organisation reaches the point where everyone needs their own account with SSO and audit logs, that is the conversation to have about Enterprise.
Their answers autosave as they go, so a network blip or a closed tab doesn't wipe their progress — if they come back within seven days, the survey resumes where they left off. Drafts don't count toward your response total, but they stay visible on the Responses tab if you want to see where people drop off. Nothing is deleted.
You will have seen it coming — the account page shows current usage against each allowance throughout the month. New runs of that one tool pause until the reset; everything else, including your surveys, your live fielding, and all existing results, carries on untouched. If you are regularly running into a ceiling, that is worth a conversation rather than a workaround.
Start on Research, or talk to us first if you want to walk through the employee-listening setup.