When we onboarded our first set of beta customers, we were excited to showcase how our synthetic data generation tool works, help them generate data with AI, and gather feedback before going GA. However, it wasn’t long before one customer dropped out of beta, citing that the tool was too difficult to use.

This feedback was a wake-up call for our team. We soon realized that this challenge wasn’t unique to us. In fact, Gartner Peer Insights revealed that for every single vendor, the most common dislikes were about the complexity of setting up and getting started, which validated what we were hearing from our own customers. Determined to address this, our team focused on simplifying the generation process. We introduced features like smart database sizing, automated check constraint generation, and an "AI everything" setting to enable users to get started with no manual configuration.

Through this process, we learned valuable lessons about simplifying generation setup: never assume that the user has perfect knowledge of their database, never sign an issue off as solved just because it can be handled by manual configurations, and always work closely with users to learn how they use your tool. In this talk, we’ll share what worked, what didn’t, and the insights we gained while testing our simplified setup—a crucial step in our roadmap to going GA.