AI Calling; Please HoldAI Calling; Please Hold

AI has commandeered the enterprise communications conversation. It’s been true for some time now, but especially for the past three years, AI worked its way across all of the enterprise communications modalities. Examples of some groundbreaking capabilities include, transcripts and translations, suggested replies, improved noise suppression, and, of course, summarization. As a result of these AI-powered capabilities, how we interact, meet and collaborate are all changing — mostly for the better.

However, AI has not had much impact on how we make and receive calls. The basic calling experience hasn’t changed that much. It’s surprising because calling is such a core feature of enterprise communications, and so many of the UCaaS providers are focused on AI. I had come to the erroneous conclusion that perhaps there are not many ways to change or improve the basic calling experience. Then I got a Pixel smartphone from Google.

Smartphones start with the basic choice between iOS and Android. Apple iOS is limited to Apple iPhones. Google Android is available on many brands. Azus, Motorola, OnePlus, and Samsung all have slightly different takes on Android. Google’s uses its Pixel phones to showcase what Android can do. Last year, both Apple and Google turned their phone-update-focus toward AI-enhanced features.

The Pixel phone includes a year of Google Gemini and offers a variety of AI-inspired capabilities, but what I found most intriguing are its features that improve calling -- features I have not seen on UCaaS clients. That’s an accomplishment that deserves recognition: Google Pixel phones have AI features that have not been rapidly copied. That’s rare in enterprise communications. Google has an advantage over most UCaaS providers, control over its hardware. However, most of the features could be replicated on softphones and proprietary hard phones.

Google’s AI Features that Improve the Calling Experience

Let’s start with Call Notes. This feature saves important details from phone calls. Call Notes provides a private summary of the conversation after a call, including captures spoken details such as dates, addresses, and phone numbers. The notes are stored in the call log. It’s similar to some of the advanced features we have seen in meetings but it’s available on PSTN calls. As with meetings, if activated, it will alert all parties automatically at the start of the call. What’s particularly interesting is that the AI is running on-device for improved privacy.

Next up is Direct My Call. This is one of my favorite features. It’s a visual IVR. The Pixel detects, hears, and displays auto attendant menus. For example, if the recording says, “press one for new orders or returns,” a virtual button is created on the screen that’s labeled “New Orders Or Returns.” In terms of AI, this is low hanging fruit. I have seen the more common intermediate step that directs callers to say a number instead of dialing it, , but why are we using digits for routing at all in 2025?

Pixel phones also have virtual hold. This feature is typically associated with call centers not end-user phones, but it makes a lot of sense as a calling feature. On the Pixel, it’s called Hold for Me. It must be turned on in the phone app’s settings and it is limited to certain countries. The Pixel detects when the person making a call gets stuck on hold, then prompts the caller with its Hold for Me feature. If the user accepts the feature, the Pixel will ring to alert them when the hold time is over. This feature is restricted to mostly English-speaking countries and, for some reason, to toll-free numbers.

Virtual Hold is best implemented bilaterally. For example, with Verint, both the contact center and the caller agree to use the feature. The systems typically disconnect, and the caller gets requeued in a shorter line when their wait period ends. This bilateral approach sets expectations on both ends. Google’s implementation on the Pixel is unilateral, so the contact center isn’t involved, nor is the call disconnected. The Pixel monitors for a live person to return and then alerts the caller with a ringtone. I don’t expect much patience should there be a delay between the alert and the caller returning to the phone.

Clear Calling has been available since the Pixel 7. It’s essentially a noise-cancellation feature optimized for speech. It reduces background noise regardless of which end of the call it comes from. Its noise cancellation can also improve the experience of a a low-quality speakerphone. The feature must be turned on manually in the Pixel’s settings, so many Pixel users are likely unaware of the feature. Once activated, it’s on for all calls.

The Pixel has a call screening option. Google calls it “Call Screen.” The feature can be activated in Maximum, Medium, and Basic levels of privacy: Maximum blocks any number that isn’t in your contacts, Medium uses Google algorithms to block “suspicious” numbers, and Basic blocks known spammers.

Some other notable features include Read CallerID to have the phone announce who’s calling. I also like the option “virbrate first then ring gradually.” That’s the name and description and solves the problem of the ringer being too loud in quiet environments and too soft in noisy environments (though detecting the ambient noise level could also work). 

I love what Google is doing with AI for calls, but I must point out that Google makes many of these features complex to discover and activate. Some are activated in the phone’s general settings (Clear Calling), some are in the phone app’s settings (Hold for Me), many are turned off by default, and in some cases the feature are completely hidden (as opposed to disabled if if they unavailable due to location. Features I have in the US are completely missing from menus when I travel abroad.

The Pixel phone also has some advanced AI tricks that I don’t understand at all. For example, it has a built-in feature to suggest the best time to call a business. It’s like how Google Search can show how busy a business is during the day. When you enter a business phone number in the dialer, the phone will display average wait times by day and hour if available. I’ve always assumed the physical store traffic is determined by the number of Android phones in a given location. Wait time is more complex because it’s impacted by both the demand/load as well as staffing. Presumably, Google is estimating this by obtaining historical wait times from somewhere.

Another advanced feature that I rarely use is the complete real-time call transcription feature. This is tied to the Pixel’s Live Caption system intended for hearing-impaired users. It’s been available since the Pixel 2, and displays real-time captions in one of multiple languages (and can translate). I have used the feature to get passed IVRs in foreign countries, but it takes over the display, so navigating menus or accessing anything on the phone during a call becomes difficult.

UCaaS providers won’t be able to implement features that use the display on industry-standard SIP phones. I would love to see these features become more commonly available on UCaaS softphones (PC or mobile). It won’t be practical to implement visual features on a SIP phone, but some providers support proprietary phones. Avaya, Microsoft, and Mitel could implement them on their desktop phones. It would be nice to see these features on UCaaS, particularly since most providers tout their AI prowess. Maybe 2025 is the year it happens.

Dave Michels is a contributing editor and analyst at TalkingPointz. Innovations in text? Apply for the Innovation Showcast at Enterprise Connect 2025. Details here.