# An interviewer that asks why, with everyone at once.

> An AI interviewer that collects feedback in a short conversation: it asks a follow-up when an answer stays vague, in each respondent's own language, and sums up in their words.

Web page: https://febaso.com/features/ai-interviewer · Updated: 2026-10-08

Every answer gets the question a good researcher would ask. In about four minutes.

- Asks a follow-up when an answer stays vague
- In their own language, typed or spoken
- Stops when someone wants to stop

## From a vague answer to what someone means.

1. **You choose what matters.** Per question you decide whether Febaso asks a follow-up: never, once, or up to three times.
2. **Febaso asks a follow-up.** The moment an answer stays vague, with the respondent's own words above it.
3. **Everyone hears what you did.** Whoever wants to gets a message when something changes, in their own language.

## A good interviewer, in rules.

- **Follow-ups.** Per question none, one, up to two or up to three. After two short answers in a row, it stops asking follow-ups.
- **Stopping.** Whoever wants to stop can always stop, and every question can be skipped.
- **Length.** About four minutes. At that point it asks once whether someone wants to wrap up.
- **Languages.** Everyone in their own language. You read it all in yours, with the original next to it.
- **How.** Typed or spoken, on a phone or a computer. One link or QR code, no app or login.
- **Summary.** At the end: in short, you said this. The respondent can add to it or correct it.
- **Open about AI.** The first message says you are talking to an AI interviewer.
- **Who said what.** The organisation never sees who said what. Updates go through Febaso, apart from the answers.

## Take it yourself, and watch it ask why.

Make an interview, answer it as a customer and give a vague answer on purpose. Two minutes, with made-up data.

### What one round yields (Example data)

Northwind Summit 2026 · Northwind. Northwind, Marieke and every answer here are made up for this demo.

| Theme | Raised by | In their words |
| --- | --- | --- |
| Sessions: too many overlap | 186 attendees | "Two talks I wanted ran at the same time. I had to choose, and I picked wrong." |
| Programme: published late | 23 attendees | "The schedule came out two days before." |
| Lunch: long queues | 142 attendees | "Half an hour in line for lunch." |
| Networking: too little time | 97 attendees | "Great people, too little time to talk." |
| Speakers: worth the trip | 88 attendees | "The keynote alone was worth the trip." |

## A form records. A conversation understands.

| A form | The AI interviewer |
| --- | --- |
| "Fine" in the open field, and that's it. | A follow-up until you know what fine means. |
| The next question on the list, whatever someone answered. | A question that responds to the answer. |
| The same questions for everyone, even when they don't fit. | Follow-ups on what someone brings up. |
| One language for everyone. | Everyone in their own language. |
| After sending, you never hear back. | A message when something changes, for whoever wants it. |

## One interviewer, every question.

- [NPS](https://febaso.com/use-cases/nps.md): The score, and why
- [Customer satisfaction (CSAT)](https://febaso.com/use-cases/customer-satisfaction-csat.md): CSAT with follow-up questions
- [Employee engagement (eNPS)](https://febaso.com/use-cases/employee-engagement-enps.md): Engagement and eNPS, anonymous
- [Member feedback](https://febaso.com/use-cases/member-feedback.md): For associations
- [Course evaluation](https://febaso.com/use-cases/course-evaluation.md): Courses and classes

## Privacy and trust

- **Processing in the EU.** And never used to train AI models.
- **Anonymous when you choose.** You choose per interview whether answers are anonymous. Anonymous means you never see who said what.
- **Open about the AI.** Everyone knows from the first message that it is an AI.
- **Traceable.** Every theme shows the count and the verbatim quotes.

[Read how we handle privacy](https://febaso.com/privacy)

## Frequently asked questions

### Do people know they are talking to an AI?

Yes. The first message says you are talking to an AI interviewer, and that stays the case. That is also how it should be under the EU AI Act.

### Can the AI make things up in the summary?

The summary is in the respondent's own words, and they can add to it or correct it. You see the verbatim quotes for every theme, so every conclusion can be traced to the answer itself.

### Can I decide what it asks?

Yes. You approve the questions, and per question you choose how many follow-ups are allowed. What Febaso may only know as background, it never says out loud.

### How long does a conversation take?

About four minutes. Febaso estimates the length from the questions and follow-ups, and at that point asks once whether someone wants to wrap up. Whoever wants to stop can always stop.

### Which languages does it work in?

Everyone gets the questions in their own language and answers by typing or speaking. A Polish and a Dutch answer about the same problem end up in one theme, and you read it all in your language, with the original next to it.

### What happens to the answers?

They are processed within the EU and never used to train AI models. Every customer signs a data processing agreement, and data is deleted on request.

## Let it hold the first conversation.

Set up your first interview yourself, for free, and try the conversation yourself first. Rather talk first? Book an intro.

- [Book an intro](https://febaso.com/features/ai-interviewer#contact)