Pick the pictures.
Two things. First, the captions are now a short bold benefit header plus a normal-weight explanation (for example "Finally know what's in your storage unit", then how it works) instead of one solid block of bold. Second, below the full onboarding is a Pick the pictures gallery: twenty photoreal scene options across storage, moving, insurance, documents, kids and garage, each with suggested labels, so you can choose which image goes in each use case. Scenes and labels are tuned to what each buyer persona said would sell them. No em dashes.
The full flow
swipe / click / arrow keys · on AI consent, tap Learn more to see the explainer, then Back or Allow
Pick the pictures
Twenty photoreal scene options, each with suggested labels tuned to what each buyer persona said would sell them. Tell me which one goes in each use case (and which labels), and I will wire it into the flow above with the bubbles positioned on the items.
labels here show the payoff (where / when / value), not just the item name
The search highlighter, in motion
Both looping so you can see them move. Same result, same thicker lines, same pills. The only difference is how the typing itself is called out.
Two-tone chip
The query in a matching two-tone pill. Consistent with the results.
Focus ring
Just a soft glow on the field. Quieter, lets the results lead.
pick one and it is the last decision