Product

Iris 2.1 adds voice to intake

Iris 2.1, now with voice
The 2.1 release film. A visitor taps the mic, talks through a visa denial, and the firm gets the lead at the end.

Iris 2.1 is live for every firm running Iris. Visitors can talk to her instead of typing, and the conversation they have out loud still arrives as the same structured lead.

What changed in the conversation

The mic button now opens a conversation instead of a recording. One tap, Iris greets the visitor out loud, and the mic stays open. There is nothing to press between turns.

Answers come back in about a second. That number is measured from the visitor's last word to Iris's first, end to end, on a normal connection. It is short enough that people keep talking instead of waiting for a machine to catch up.

Interruption works the way it does with a person. Talk over her and she stops mid-sentence, then answers what was actually asked. Nothing she had already said keeps playing underneath you.

English and Spanish are both spoken. If a caller starts in one and switches, Iris hears the switch, changes her spoken voice, and keeps the case where it was. Those are the two languages 2.1 speaks aloud.

What the firm gets

The same lead. Name, phone number, practice area, the story, the score, and the reason behind the score. Iris asks for the details during the conversation rather than at a form at the end, and the capture arrives in the same shape a typed intake does.

The same transcript. Each spoken turn is written into the conversation as text, with timestamps, so the morning summary and the full record read like an intake that was typed. The record does not change shape because the visitor spoke.

Nothing to install. The update is on our side. Firms running Iris did not touch their site, their snippet, or their settings.

What it does not do

It does not answer the phone. Voice runs inside the widget on the firm's site, so the visitor has to be on the site to use it. There is no number to forward and nothing to port. A firm that wants Iris answering inbound calls is asking for something we have not built.

Voice also needs microphone permission, which the browser asks for on the first tap. Anyone who would rather not talk can keep typing. Both paths end in the same lead.

The record is the transcript. Speaking does not turn an intake into a phone call.

The product page now leads with 2.1 and keeps the 2.0 material below it, as the foundation it was built on. The bilingual page covers how the Spanish conversation is written rather than translated.

Share this article