Talk to the Agent

On the Agent page, the mic in the message box lets you speak a request instead of typing it. Press it, say “add a status column to orders and show it on the list”, and press it again. Your words appear in the message box.

What happens when I stop?

What you said lands in the message box, after anything you had already typed. Nothing is sent yet: add to it, paste in an error message or a link, fix a word, then send it as you would any message. You can record again to add more.

How do I record?

• Mic starts recording. It turns to the accent colour, with a meter of your microphone beside it. The message box stays as it is, and you can keep typing in it.
• Press the mic again to stop. The words go into the message box, after what is already there. Don’t want them? Delete them there, like any text.

A recording can be up to two minutes long. At two minutes it stops by itself, as if you had pressed the mic.

If nothing was heard, the message box is left as it was and the page says so.

What happens to my voice?

• Your audio goes from your browser straight to Amazon Transcribe in the EU, which turns it into text. Sapilon never receives or stores the recording.
• Only the text you send is kept, like anything you type to the Agent.

The first time you use it, a short sheet says the same before the browser asks for the microphone.

What does it cost?

Each recording is billed to your organization like other AI use: one small flat charge per recording, next to the Agent turn your message starts, which is billed as usual. If your organization is out of credit, recording does not start.

It does not work

• “Microphone access is blocked.” Allow the microphone for this site in your browser’s site settings (the icon next to the address), then press the mic again.
• No mic button. Voice is not switched on for your account yet.
• The words are wrong. Fix them in the message box before you send, or speak a little slower and closer to the microphone.
• “Too many voice messages this hour.” Wait a while, or type the request.

Voice input works in current Chrome, Edge, Safari (on Mac and iPhone) and Firefox. It is transcribed in your browser’s language when Amazon Transcribe supports it, and in English otherwise.