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

# How to Create a ListenLand AI Interview From Scratch

> Step-by-step guide to building a ListenLand interview — from writing questions to configuring AI behavior, voice, and access settings.

Creating an interview is how you tell the AI what to ask and how to behave. Walk through every setting so your interview captures exactly what you need.

<Steps>
  <Step title="Open the dashboard">
    Sign in at [listen.land](https://listen.land). Click **New Interview** in the top right of your dashboard.
  </Step>

  <Step title="Write your title and overview">
    The **title** is for your reference only — respondents never see it. It labels the interview in your dashboard.

    The **overview** is the most important field. It's read by the AI as background context before every conversation begins. Describe your research goal and any important context about your respondents. The AI uses this to frame its questions, interpret answers, and decide what to probe on — even when respondents go off-script.
  </Step>

  <Step title="Add questions">
    Click **Add Question** to add each question you want the AI to cover. Questions are private to the AI — respondents never see a list. The AI weaves them naturally into the conversation rather than reading them aloud as a script.

    For each question, set a **follow-up depth**:

    | Depth      | Behavior                                                                                                                                    |
    | ---------- | ------------------------------------------------------------------------------------------------------------------------------------------- |
    | **Low**    | The AI asks the question and accepts the first answer, then moves on. Best for factual or demographic questions.                            |
    | **Medium** | The AI probes once or twice if the answer is brief or surface-level. Good for general topics.                                               |
    | **High**   | The AI digs deep with multiple follow-ups — exploring specifics, examples, frequency, and feeling. Best for the questions that matter most. |

    You can also mark individual questions as **required**. The AI won't end the conversation until every required question has been substantively addressed.
  </Step>

  <Step title="Set the interview mode">
    Choose how respondents will interact with the AI:

    * **Voice** (default) — Respondents speak aloud. Conversations are recorded and you get audio back alongside the transcript.
    * **Text** — Respondents type their answers. Good for async or written feedback.

    The mode you select becomes the default when respondents land on your interview. You can configure this per interview.
  </Step>

  <Step title="Configure access">
    Choose who can respond:

    * **Open** — Anyone with the interview link can start a conversation. Best for public feedback, community research, and open surveys.
    * **Access Code** — Respondents must enter a code before they can start. Best for private panels, paid research, or internal surveys. Set the code respondents will need to enter, or copy the auto-generated one.

    See [Access Control](/guides/access-control) for a full breakdown of both modes.
  </Step>

  <Step title="Customize the AI persona (optional)">
    Two optional fields let you tune the AI's behavior beyond the default:

    * **Persona instructions** — Free-text instructions that steer the AI's tone and style. For example: *"Be warm and encouraging"*, *"Keep responses concise and direct"*, or *"Don't accept abstract answers — always ask for a specific example."* The AI's core structure (one question at a time, depth rules, ending logic) is platform-controlled; persona instructions only shape voice and emphasis.
    * **Language** — The language the AI should respond in. Defaults to English. You can enter any language in natural terms, e.g. `French`, `polite Japanese`, or `Brazilian Portuguese`.
    * **Voice** (voice mode only) — Choose from a range of AI voice options to match the tone of your research.
  </Step>

  <Step title="Publish the interview">
    When you're ready to start collecting responses, toggle the interview status to **Active**. The interview link is now live and accepting conversations.

    You can pause the interview at any time to temporarily stop new conversations, or close it entirely when your research is complete.
  </Step>
</Steps>

<Tip>
  Write your overview as if you're briefing a human interviewer. The more context you give the AI, the more relevant its follow-up questions will be. Include your research goal, who the respondents are, and any nuances the AI should keep in mind.
</Tip>

<Warning>
  Once respondents start responding, avoid making major changes to questions or the overview — it can create inconsistency across responses. Changing the questions mid-study means earlier and later conversations were guided by different criteria, making synthesis harder.
</Warning>

***

<Card title="Share your interview" icon="link" href="/guides/share-interview">
  Your interview is live — now get it in front of respondents. Learn how to share your link across email, Slack, surveys, and more.
</Card>