The project started with high ambiguity and no clear requirements, so I worked through informed assumptions, rapid explorations, and continuous alignment with product and engineering to shape an initial direction.

Rather than designing a fully defined AI feature upfront, I focused on creating a flexible foundation that could evolve with both the product and AI capabilities. I framed the experience around three key user moments: when users don’t know how to start, when they are analyzing an existing funnel, and when they want to add a new step but lack context.

Exploration began by reviewing AI interaction patterns across tools to understand mental models, followed by iterative prototyping in Figma within a key constraint: reusing Datadog’s existing Bits AI side panel.

A major focus of the exploration was defining when AI should appear, how suggestions should be applied (preview vs direct), and how to maintain reversibility and user control within analytical workflows.

Due to the fast timeline, validation relied on continuous design critiques, scenario walkthroughs, and cross-functional feedback. I also incorporated Figma AI and Cursor into my design process to accelerate exploration, prototype interaction behaviors, and refine guidance patterns within realistic workflows.