Xiaoju Survey is an open-source platform for building and running online forms, surveys, quizzes, and polls. It was created by DiDi, the Chinese ride-sharing company, and is designed for both individual users and enterprise teams who need to collect and analyze data at scale. With this tool, you can create a form using more than 40 different question types, including text inputs, multiple choice, rating scales, file uploads, and voting options. The editor supports conditional logic, meaning you can show or hide questions based on how a respondent answers earlier ones. You can also reference answers from one question inside another, which is useful for more complex surveys or assessments. Once responses come in, the platform provides built-in reporting: per-question statistics, cross-tabulation (comparing how different groups answered), and multi-channel breakdowns. The README describes the system as having been tested at very large scale internally at DiDi before being released publicly. A notable feature is AI-assisted form creation: you can describe what you want in plain text and the system will generate a questionnaire draft for you. This is connected to a large language model and is built into the editor. The tech stack is Vue 3 on the front end and NestJS with MongoDB on the back end. Deployment is handled via Docker, and two image variants are available: a slim production image and a fuller image that includes debugging tools. The project also ships an SDK for embedding small survey widgets inside other apps. Permissions and team collaboration are supported, with workspace management and role-based access control. The README mentions data encryption in transit, sensitive-word filtering, and anti-ballot-stuffing measures as security features. The project is written in TypeScript and the source code is available on GitHub under an open-source license. Documentation and a hosted demo site are linked from the repository.
← didi on gitmyhub — every repo by this author, as a profile.
Verify against the repo before relying on details.