Every phone number is inherently two channels, voice and messaging, and both reach the same device in the same hand. Most voice AI, however natural it sounds, uses only one of them.
Imagine a guest who calls the front desk because the room is sweltering and the thermostat won’t respond. The AI agent apologizes, offers a different room, and asks whether it can text over the available options. Three rooms arrive in the guest’s messages app as a carousel with photos and descriptions, and the guest taps one while the call continues.
Handled by voice alone, the same call has the AI agent reading three room descriptions out loud, the guest trying to keep them straight, and the whole list read again when they lose track. There are no photos, and after the call there is nothing to look back at.
Voice is a brilliant channel with structural limits
Speaking is the most natural way most people express what they want, and it works when their hands and eyes are busy. It is a poor way to carry certain kinds of information, though. It is very hard for a customer to juggle six similar flight options in their head and remember which one they wanted. Describing damage over the phone never gets across what one photo would. And a confirmation code read over the phone has to be spelled out letter by letter, and the customer still has to write it down somewhere.
The industry’s answer has been to make the AI’s voice more human, with better turn-taking, better interruption handling, and warmer tone. That work matters, and we have done plenty of it ourselves. A more natural voice still cannot show the caller a picture.
Customers already voted for both
In Zendesk’s CX Trends 2026 research, 76% of consumers said they prefer companies that let them use text, images, and video within a single thread over ones that make them start again when they switch formats, and 74% said repeating information frustrates them. Customers want one conversation that uses both channels.
The messaging channel has also become much richer. iPhones joined Android on RCS with iOS 18 in 2024, and Google reports more than a billion RCS messages a day in the US. RCS and Apple Messages for Business both support carousels, quick replies, and pickers in the phone’s default messages app, and Apple Messages supports payment with Apple Pay on the device. The customer does not download an app or open a browser for any of it.
One conversation, two modalities
Quiq’s Voice AI treats the call and the message thread as the same conversation, so the AI agent can move any piece of the interaction to whichever channel handles it better. Choices go to the screen as a carousel or a picker. Payment happens with a tap on Apple Pay instead of a card number read to a recording (more convenient and more secure). Dense details get confirmed aloud in one line (“you’re confirmed for Tuesday, I just texted the full itinerary”) while the exact dates, times, and addresses arrive as a message the customer keeps. And if the answer is something physical, such as the damage or the serial plate on a unit, the customer photographs it and texts it back during the call, and the AI agent acts on the photo while they are still talking.
The customer replies however they like, by tapping the second room or by saying “the second one.” Nothing has to be repeated, because both channels share one conversation history.
This works because Quiq’s platform grew up on messaging, where a conversation already had to outlive any single session. When we added voice, a phone call became another way into that persistent conversation, and sending or receiving a message mid-call did not need a second system. The same design lets a human agent on a live call send the room carousel with one click while the AI handles the message mechanics.
