> ## Documentation Index
> Fetch the complete documentation index at: https://docs.trypillow.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Interview guide

> How your interview guide decides what your agent asks: scripted questions and exploratory topics.

Your **interview guide** is the ordered list your Researcher Agent works through in every interview. Its items come in two kinds, and they work at different levels: a **Question** is the literal wording to send, while a **Topic** is an area you want explored, left for your agent to phrase.

## Questions and topics

| Item         | What your agent does                                                                             | Best for                                            |
| ------------ | ------------------------------------------------------------------------------------------------ | --------------------------------------------------- |
| **Question** | Asks your wording word for word, translated faithfully if the interview runs in another language | Wording you need identical across every participant |
| **Topic**    | Writes its own questions around the area, picking the angle from each answer                     | Open-ended areas where you want unexpected themes   |

One guide can hold both, and any item can switch kind while you build. Pillow rewrites the text when it does, since the two describe different things: a topic describes an area to explore, a question states a script.

The other difference is coverage. A topic is covered the moment your agent knows the answer, so one rich reply can settle several at once. A question is only covered once it has actually been asked, even if the participant already volunteered the answer.

## Follow-ups

**Interview depth** sets how often your agent follows up on an item before moving to the next one:

| Depth        | Follow-ups per item                        |
| ------------ | ------------------------------------------ |
| **Express**  | None; every message moves to the next item |
| **Quick**    | One                                        |
| **Balanced** | One, and a second on rich answers          |
| **Deep**     | Two, and three or four on rich threads     |

An item can carry its own **Follow-ups** instructions, which override this rhythm for that item alone. They steer where your agent digs, while depth still sets how much.

## The main goal

A one-sentence summary of what the study is trying to learn. The Study Builder uses it to keep your items pointed at the same outcome, while your agent interviews from the items themselves.

## Where the guide lives

You edit the guide in the Study Builder while a study is in **Draft**. Going live locks the questions, and the guide stays readable under **Settings → Brief**, where the study depth, study goals, research notes, agent, outreach message and languages remain editable. Those edits stay unpublished until you click **Save and publish** in the banner, and participants keep the published brief until then.

## Further reading

<CardGroup cols={2}>
  <Card title="Write your interview guide" icon="list-checks" href="/guides/interview-guide">
    Build your guide from questions, topics, or both.
  </Card>

  <Card title="Create a study" icon="plus" href="/guides/create-study">
    Take a study from objective to live.
  </Card>

  <Card title="Interviews" icon="messages-square" href="/concepts/interviews">
    How your agent runs each interview.
  </Card>

  <Card title="Notes" icon="notebook-pen" href="/guides/agent-notes">
    Rules your agent follows in every interview.
  </Card>
</CardGroup>