Concept
While designing an AI-powered search experience, I needed a loading spinner that visualizes the LLM’s cognitive process, both divergent and convergent. The goal was to express how the model first explores broadly, then narrows its focus to surface the most relevant information.
Process
As with many of my ideas, it started with a quick sketch. I then translated the concept into simple shapes in Figma to block out the core structure. Once the direction felt right, I moved into After Effects to experiment with animation timing.
I intentionally avoided linear motion: the final animation begins with a rapid expansion, representing the model’s divergent reach across multiple sources, then slows into a more deliberate contraction, symbolizing the convergent synthesis of information.
Once I was satisfied with the pacing, I rebuilt the animation in Cinema 4D, adding a subtle glass-like material to the circles. This gave the spinner a dimensional, transparent feel, echoing the clarity and depth of the AI’s reasoning process.
