Presence for everyone — see what works for the way you browse.
An avatar that only works for some users isn't presence — it's decoration. Keyboard, screen reader, captions, reduced motion, low power: this page says what the site and the avatar runtime do today, where we still fall short, and where to write when something blocks you. NetShow Alive builds to WCAG 2.2 AA — a target we hold ourselves to, not a certificate we claim.
Found a barrier? Tell us ↓Keyboard & focus
Site controls — theme switches, style pickers, forms, the guide — are native links, buttons and inputs you can reach by keyboard, with a visible focus ring. Every page that uses this site's header carries a skip-to-content link as its first tab stop.
Captions & transcripts
Demos are caption-first: the guide's scripted conversation writes each line into a polite live region as text, and the live agent widget carries the same kind of caption band, so what a character says is never audio-only. We don't publish videos today, so there is nothing uncaptioned to find.
Screen-reader state labels
The avatar element is exposed as an image with a text description — character, style, framing, outfit and current agent state ("Listening", "Thinking", "Speaking") — that updates whenever the state changes, so a screen reader reads the state as words instead of guessing from motion.
Reduced motion & low power
The runtime honors prefers-reduced-motion automatically: idle sway, hover, pointer-follow and state shimmer switch off. Setting power="low" caps drawing at about 24 frames a second for older devices; an avatar scrolled out of view drops to about one frame every 1.5 seconds, and a hidden tab stops it.
Not color alone
Agent states pair posture with a text label — never a color change alone. Body text on this page measures above AA contrast in all three themes. The links on this page are 44px tap targets; the three theme dots in the site header are still 36px, and that is on our list.
Accessible microphone controls
The guide runs with no microphone at all. In the live agent widget the microphone is a button with a text label, and the agent's state is written on screen as words — LISTENING, THINKING, SPEAKING, ONLINE, OFFLINE — never as a pulsing light alone.
How we handle ASCII art.
Screen readers should never be forced to read an avatar character-by-character. The ASCII render is drawn on a canvas inside an element exposed as one image, so a screen reader gets a single description ("Nova — animated human character in ASCII style, head-and-shoulders framing. Agent state: Listening.") rather than the characters.
If you want the characters themselves, the SDK's getAsciiText() returns the current frame as real text you can copy, paste and style.
Found a barrier?
Accessibility issues are fixed with the same priority as functional bugs. Write to ai@netshow.ai — include your assistive tech and browser if you can, and tell us which page.