What taking part in an AI-moderated interview is like
By Hey Juno · Published · Updated
Taking part in an AI-moderated qualitative interview with Juno, the AI researcher from Hey Juno, feels like a friendly chat in your browser. You open a link, read what the study is for, give your consent, then answer questions by typing or speaking, in your own language. You can skip a question you would rather not answer, take a break, or stop.
This guide is for two groups: teams planning customer research who want to know what the people they invite will go through, and the people invited. Juno is the AI researcher from Hey Juno.
What happens when you open the link
Participants arrive through an invite link. They see the study introduction, then a consent step, then any pre-interview questions, then the interview itself. The whole thing runs in the browser. There is nothing to download or install.
The interview uses the study's brand, so a participant sees the name of the team that invited them. It reads like a chat, or sounds like a conversation in a voice study, so people know what to do straight away.
Everyone who opens the study link completes the study's introduction, consent and any pre-interview questions.
Consent and what you are told
Before the interview starts, participants are told what the research is for and how their answers will be used. The standard Hey Juno works to is that answers are used for the research they were given for, and are not quietly repurposed for something else.
Participants can withdraw. They can also ask for access to what they shared, for it to be corrected or for it to be deleted. To use these rights, ask the team that invited you.
If you are planning a study, write the introduction in plain words: who you are, what decision the research informs and what you will do with the answers.
Text or voice
Every study runs as a text or a voice interview, and the questions are the same. In a text interview, people type in a chat whenever it suits them. On supported pages and languages, they can dictate instead of typing. In a voice interview, Juno talks with the person, hears how they say things and asks follow-ups in the moment. Voice is available on some plans.
Text suits people who want to think before they answer. Voice takes a little more from participants, but voice tends to give fuller answers.
Your own language
Juno interviews people in the language set for the study, so they can answer in words that feel natural to them.
What the interview is like
Juno follows what you say. It has research objectives for the study, not a fixed script, so it asks about your reasons, your examples and what you meant by a word. It will follow a point that matters to you, not march through a list.
The interview can also include rating scales, ranking, yes or no and single-choice questions, and some studies show images, video, audio or links.
The team wants your experience and your reasoning, including when it differs from what they expect. If an answer is very thin or off topic, Juno may gently invite you to say a little more.
How long it takes depends on how the team set up the study. A countdown timer is shown by default, so you can see roughly how much time is left and the interview never feels open-ended.
Declining, pausing and stopping
You are never obliged to answer. When someone cannot or will not answer a question, Juno finds another way in or moves on rather than pressing. Hey Juno's standards also rule out deception and pressure, and it will not coax out information a study has no reason to collect.
If life interrupts, you can leave and come back from the same browser within the comeback window for the study. The window is four hours for studies on qualified-interview credits and 24 hours for other studies, unless a different window is set. Answers already sent stay attached to the interview, so you do not start again. You can stop at any time.
Privacy and redaction
Juno captures what the study needs and no more. Transcripts and quotes are not automatically redacted today. Juno plans to mask hard identifiers such as full names, email addresses and phone numbers in text, but that is not yet switched on, so assume the team sees what you say as you say it. Access is limited to people in the team's organisation who can see the study. Where a study records audio, you agree to the recording before the interview starts, and the team can listen to the original recording. The team sets how long answers are kept and can have them deleted.
In analysis, identity is carried only where the research genuinely needs it. Health, finances and family details get tighter handling than ordinary answers.
If someone you mention comes up in the conversation, Juno treats them as background. It does not profile or pursue them as a subject.
Wellbeing safeguards
Most interviews are ordinary conversations. Juno applies a wellbeing framework in every interview, and the protections cannot be switched off. In the rare case of present self-harm or imminent danger, a safety rule ends the session, shows crisis guidance and alerts a person. Juno is not an emergency service. If you are in danger, contact local emergency services.
The full detail is in participant wellbeing in AI-moderated research and Hey Juno's safety standards.
Why a good participant experience produces better data
A person who feels pressed or exposed does not give slightly worse answers. They give the acceptable answer, the short one, or they leave. The conditions that make it safe to speak are the conditions that make the answer honest.
That is why the design choices above are not decoration. Clear consent tells people why their words matter. Being able to decline means the answers you do get are willing ones. Their own language means they do not have to translate their thinking before they speak. Pausing and resuming means a busy person can finish rather than drop out. And an interviewer that listens and follows up gives people room to explain what they actually think.
For teams, the practical test is to take the interview yourself before launch. Juno lets you preview a study and take it yourself (see real participants versus synthetic respondents for how rehearsal differs from research). Ask whether you would be comfortable answering it honestly.
What this guide doesn't cover
This guide describes the standard Juno experience. Studies can be set up differently, so a study's own introduction takes precedence. It does not give interview lengths, because they are set per study. It does not cover payment or incentives, which depend on the team that invited you. It is not legal advice on consent in your country.
Frequently asked questions
How long does a Juno interview take?
It depends on the study. The team sets the length, and a countdown timer is shown by default so you can see roughly how much time is left.
Can I answer by voice instead of typing?
Some studies run as voice interviews, where you speak with Juno. In text studies, supported pages and languages let you dictate instead of typing. Check the study's invitation to see which applies.
Do I have to answer every question?
No. If you cannot or do not want to answer, Juno looks for another way in or moves on. You can also pause and return from the same browser within the study's window, or stop.
Who can see what I say?
The team that invited you sees the transcripts and, where the study records it, can hear the audio. Transcripts are not automatically redacted yet, and access is limited to people in the team's organisation who can see the study. The team sets how long answers are kept, and you can ask for yours to be deleted.
Is this a person or an AI?
An AI. Juno is an AI researcher. It is built to listen and ask good follow-up questions, and the standards it follows were set and are reviewed by qualitative practitioners with more than 25 years of fieldwork.