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

# AI Dynamic Questions

AI Dynamic Questions, powered by the Field Agent, automatically generate intelligent open-ended questions based on respondent input, so every survey adapts in real time without manual logic.

<img align="center" src="https://mintcdn.com/sprig/_1uWncsTBTa1sJbM/images/ed53d78f7401103e90ef7dc573261f1ab7231dff26ba1029e46ce73fe8d5a363-Changelog-AI_Dynamic_Questions_1.png?fit=max&auto=format&n=_1uWncsTBTa1sJbM&q=85&s=a0f0c07c3c0d793491a870efad5e3d5b" width="1800" height="1284" data-path="images/ed53d78f7401103e90ef7dc573261f1ab7231dff26ba1029e46ce73fe8d5a363-Changelog-AI_Dynamic_Questions_1.png" />

## How it works

* **For the study designer**: Insert an AI Dynamic Question anywhere in your study following an open text, rating, or multiple choice question.
  * You can optionally provide a custom prompt to guide each question.
  * The system always respects skip logic and quotas.
  * There is no limit on AI dynamic questions that can be added to any study
* **For the user**: The AI generates a concise, neutral question in your study’s primary language. If it can’t generate safely or quickly, the question is skipped.
  * Moderation filters are active to block PII, bias, and unsafe content.
* **For the analyzer**: AI-generated questions and their corresponding responses are available in the Reports tab. They appear alongside your standard results, ready to review individually or together.

<Note>
  #### Best Practice

  Start with one or two AI Dynamic Questions per survey to gain depth without adding respondent fatigue.
</Note>

## Why it matters

* **Dynamic probing**: Enables deep questioning based on all prior responses and context signals.
* **Reduced Setup Time**: Eliminates the need for complex branching logic needed, since the AI handles dynamic questioning automatically.
* **Faster analysis:** AI responses are grouped and displayed so researchers can quickly identify patterns & themes.

## How the questions are generated

The Field Agent utilizes a variety of signals to ensure questions remain relevant and aligned with research objectives.

<Accordion title="Context Signals">
  * Team context (if available, *coming soon!*)
  * Agent Context (if available) to stay aligned with your research objective
  * All prior questions and responses in the survey
  * Question prompt (if available)
  * NLP signals such as:
    * intent (e.g. pricing, bugs, feature requests, churn risk, compliments),
    * sentiment,
    * topics (e.g. onboarding, billing, performance, UX),
    * and entities (e.g. product, competitor, plan tier)
</Accordion>

<Accordion title="Question Styles">
  * Clarification: Resolve ambiguity or vague answers
  * Depth: Probe for reasons, mechanisms, or examples
  * Context: Compare with past experiences or alternatives
  * Impact: Explore effects, severity, or frequency
  * Sentiment: Acknowledge tone and ask what influenced it
  * Keyword: Drill into mentioned keywords or entities
</Accordion>

<Warning>
  AI Dynamic Questions are available on Enterprise plans only, and can be added to In-Product Surveys, Long-Form Surveys, and Feedback Studies.
</Warning>

## Prompt Examples

There are endless ways to configure AI Dynamic Question prompts to ensure high quality questions are asked to your respondents. A few common prompts are listed below.

<Note>
  You can test the AI output at any time in preview mode prior to launching your study.
</Note>

1. Based on everything the respondent has shared so far, ask one question that fills the biggest gap in understanding their \[workflow, feedback, experience].
2. If the respondent has mentioned a frustration or pain point earlier in the survey, probe deeper on that. **Do not** ask about pricing or budget.
3. If the respondent's answers seem to contradict each other, ask a clarifying question to reconcile the difference. **Do not** point out the contradiction directly or make them feel called out.
4. Ask about a use case or scenario the respondent has not yet covered, given what they have shared. **Do not** ask demographic questions or anything about their job title.
5. If the respondent mentioned a competitor or alternative product earlier, ask what would make them switch fully. **Do not** ask for negative feedback about the competitor by **name**.
6. Based on the goals or outcomes the respondent described earlier in the survey, ask what success would look like for them in the next 3 months. **Do not** ask about specific metrics or numbers.