Simple, minute-based pricing

Surveys and chatbots use text, which leads to short, shallow answers. PopsicleAI conducts real voice conversations — the AI listens, adapts follow-up questions based on what participants say, and probes deeper into interesting responses. Participants speak naturally, which surfaces richer insights you'd never get from a text box.

For structured and semi-structured interviews, yes — and in some ways better. The AI never gets tired, never leads the witness, never forgets to ask a question, and is available 24/7. It follows your research protocol perfectly every time. For deeply exploratory or sensitive topics, a human moderator may still be preferable.

On a plan, interviews keep going at $1.00/minute (billed per second) on your next invoice, up to a monthly cap you set — 1× your plan price by default, so there are no surprise bills. Turn overage off and new participants see that the study isn't accepting sessions until you add minutes. Active sessions are never cut short: if your balance can only cover part of an interview, that participant's session is set to the time you have left before it starts (5-minute minimum).

Voice interviews are billed per second, from when the session starts to when the participant leaves. Sessions under a minute are free, and so are sessions that fail on our side or where the participant never speaks. A chat-only interview uses a flat 4 minutes however long it runs, and is free if the participant answers fewer than 3 times. Unused plan minutes roll over for one month.

Yes. Add a card and get 15 free interview minutes to use within 30 days — you won't be charged. Prefer not to subscribe? Minute packs (30, 100 or 300 minutes) are valid for 12 months.

Yes. Participants join via a simple link — no downloads, no accounts, no apps. It works in any modern browser on desktop or mobile. Voice interviews need a microphone; chat interviews don't.

Yes. In a voice interview participants can type any answer in the chat. You can also set a study to chat only: participants get a full-width chat and no microphone is needed. A chat interview uses a flat 4 minutes however long it runs, and is free if the participant answers fewer than 3 times.

Each study has agent instructions where you define the research goals, topics to explore, and specific questions. The AI uses these as a guide while adapting naturally to participant responses. You can add images, videos, ads or prototype links that the AI shows on the participant's screen when your instructions call for them, and embed a website for usability testing. You also set participant instructions and the max session length.

Yes, in 28 languages. Run a study in one language, or offer several and let each participant pick theirs before the call. The AI interviews and transcribes in that language, and the participant's screens are translated. Your instructions can stay in English.

Recordings and transcripts are stored securely and only accessible to you and the teammates you invite to your workspace. Participant data is never shared with anyone outside it. You can delete recordings at any time.

Yes, and seats are free. Invite teammates by email from Settings → Team. Everyone shares your studies, recordings, AI findings and interview minutes, so there’s one balance to top up. Teams can have up to 25 people, and only the account owner manages billing.

Yes. PopsicleAI has an MCP integration — just run npx popsicle-ai-mcp and connect it to your AI assistant. You can create studies, pull transcripts, and analyze research data without leaving your workflow.

The AI introduces itself as a research assistant. We recommend being transparent with participants that the interview is AI-conducted — this is both ethical and, in our experience, doesn't reduce response quality. Participants tend to be more honest with AI interviewers.