Sprint #35
June 22 → July 6
✅ New features
• AI Coding: Verbatim responses are now coded at two levels: codes and sub-codes. The results reflect this hierarchy, allowing for a more granular analysis of open-ended responses.
• AI Agent: A conversational agent is now available to query study results using natural language. The agent can access questionnaire questions, flat results, and cross-tabulations to provide instant answers.

• MCP: A poll&roll MCP server is now accessible and, like the AI agent, allows you to query study results directly from various external AI tools.
• User Guide: Documentation for the application, API, and MCP is now online. It can be accessed via the icon ? in the navigation bar.
🛠️ Améliorations
• Word Questionnaire Import: Blocked imports are automatically detected and deleted.
• Word Questionnaire Import: An ongoing questionnaire import can now be manually canceled.
• Hidden Variables: Multiple-choice questions are now accepted as hidden variables.
🐞 Bugs corrigés
• Respondent Questionnaire: For numerical questions with decimals, responses were displayed multiplied by 100.
🔮 À venir dans le prochain sprint
• Organization Interface: Redesigned navigation and settings organization.
• AI Agent: Direct comparison of multiple questions within the same study.