Making form creation as simple as asking

Building forms meant manually adding questions, choosing field types, and configuring responses one by one. The existing builder was flexible, but getting started could involve a lot of repetitive setup.

I wanted to explore a simpler way: what if users could describe the questions they needed, and AI could handle the initial configuration?

I used a conversational input because it is an interaction people increasingly recognize. Users can make a broad request like “Add demographic questions” or ask for just one, like “Ask about their education.” The experience works for both incremental building and generating multiple questions at once.

I considered placing AI directly in the main builder, but wanted to preserve the manual workflow for people who preferred it. A dedicated modal gives users a focused space to generate questions and lets them return to the familiar editor whenever they are ready.

These forms can collect important information, and AI can make mistakes. I chose to let users review suggested questions before adding them. Once approved, each question becomes a regular, fully editable form component.

I used Claude Code to build a functional prototype connected to an AI integration Fluid already used for other conversational features. I guided the implementation through prompts and iterated on the interactions rather than manually writing the code. Having a working prototype let me demonstrate the experience beyond static screens.

The feature is still in development. Engineering is connecting the production experience to the AI model used for internal tools and running additional testing before release. There are no adoption or impact metrics yet.