Voice-AI brand reading like an editorial magazine. Off-white canvas, Waldenburg Light at weight 300, pastel gradient orbs as the only color moments.
No saturated brand action color. Single ink-pill primary; five pastel gradient orbs as decorative atmosphere only.
Display weight stays at 300. Body Inter at +0.15-0.18px tracking — slightly looser than default for an editorial dialect.
Ink pill primary, transparent outline secondary. No saturated CTA color.
Five gradient stops drift through the page as soft radial blooms. Never used as button fills, text colors, or component backgrounds — pure atmosphere.
A horizontal card with circular play button, ink-tinted waveform glyph, and voice metadata.
Hairline-divided rows with circular voice icon, name + accent, and preview button.
Generate broadcast-grade voice from a few seconds of reference audio.
One model, every language. Native fluency without retraining.
Stream voice with sub-300ms latency for live applications.
Visual inversion signals "highlighted choice" without colored ribbons.
Small uppercase pills.
4px base, 96px section rhythm.
Pill for CTAs and badges, 16-24px for cards, 9999px for voice icons.
Hairline + soft drop. Atmospheric depth comes from gradient orbs.
Body bands, footer.
White card with 1px hairline border.
0 4px 16px rgba(0,0,0,0.04). Single shadow tier.
Atmospheric depth — never a card surface.
| Name | Width | Key Changes |
|---|---|---|
| Mobile | < 640px | Hero h1 64→32px; feature cards 1-up; nav hamburger; gradient orbs shrink. |
| Tablet | 640–1024px | Hero h1 48px; feature cards 2-up. |
| Desktop | 1024–1280px | Full hero h1 64px; feature cards 3-up. |
| Wide | > 1280px | Content caps at 1200px. |