Skip to main content
The Builder is Expected Parrot’s visual interface for creating and editing studies. Use it to build a survey or interview guide, import an existing questionnaire, or review and refine studies created by the Research Agent. Open the Builder after signing in to your Expected Parrot account.

​
Create or import a study

Start with a new study, import an existing questionnaire, or open a saved study. When importing, review the resulting questions, response options, and logic against the original questionnaire before collecting responses. You can also ask the Research Agent to draft a study, then use the Builder to edit its structured objects directly. Both use EDSL, so you can export the study to code for further customization.

​
Edit and preview

Review question wording, response types, answer options, order, and conditional logic. Configure the respondent settings for your study and preview the participant experience. Test any branching or follow-up behavior before launch. Use the Builder to make targeted changes, such as revising a scale, changing answer options, or adjusting a question in a study drafted by the Research Agent.

​
Choose your research workflow

​
Human surveys

Create a web survey for people to complete and share its respondent link with your participants. You can conduct a human study without first running a simulation. For code-based workflows and recruitment options, see Humanize and Prolific studies.

​
AI-led interviews

Create or edit an interview guide that specifies topics, prompts, and follow-up instructions. The AI interviewer follows the guide and can ask relevant follow-up questions based on a participant’s answers. Preview the interview and review the response modes supported by your study settings before sharing it with participants. The Research Agent can help draft the interview guide and analyze the resulting material.

​
AI simulations

Define AI respondent profiles, select language models, and run your study with simulated respondents. Review estimated costs and pilot the study before scaling it up. Keep simulated responses distinguishable from human observations when analyzing or reporting results.

​
Analyze and share

Review collected responses, export data for further analysis, or ask the Research Agent to help compare results and prepare a report. Use EDSL’s Results methods when you want to work programmatically. Share study materials with collaborators so they can inspect the design. Use the respondent link to invite participants to complete the study.

​
Continue in EDSL

Export studies to EDSL code when you need custom analyses, automation, or integration with other tools. Follow first steps to set up Python and the Run your first survey tutorial to learn how questions, surveys, agents, scenarios, and models fit together. For questions or feedback about the Builder, email [email protected].