The screener asks a question and waits for an answer. A dialer built before screeners existed hears a machine and either hangs up, plays a recorded message, or bridges your rep to an assistant that is still waiting to be told who is calling. In every one of those cases the phone never rings for the prospect. The dial counts as attempted, the list looks worked, and more than a quarter of your mobile reach quietly disappears.
AI call screening is an on-device assistant that picks up calls from numbers the owner has not saved, asks the caller to state their name and reason for calling, and then decides what to do: ring the owner and show what the caller said, send the call to voicemail, or hang up on likely spam. The owner sees a live transcript and chooses whether to answer. It is aimed at robocalls and scams, but it screens any unknown number, which includes every first-touch sales call.
Since iOS 26, iPhone has a Screen Unknown Callers setting with three options: Never, Ask Reason for Calling, and Silence. With Ask Reason for Calling on, the iPhone answers calls from unsaved numbers before it rings, asks the caller for their name and reason, then rings and shows the response so the owner can decide. Silence sends unknown callers straight to voicemail. Apple notes calls are not screened while roaming, and screening switches off for 24 hours after a call to emergency services.
On Pixel phones, Call Assist answers, asks who is calling and why, and shows the owner a real-time transcript. If it decides the call is spam, the phone hangs up; if not, the phone rings and shows how the caller responded. Automatic screening is available in the US, UK, Canada, Australia and Ireland, with protection levels that can screen spam, possibly faked numbers and first-time callers. Manual screening covers more countries and some other Android devices.
Screening keys off one thing: whether the number is in the owner's contacts. A cold call is, by definition, from a number the prospect has never saved, so it is exactly the call screening was built to intercept. Number reputation and local presence still matter for whether the call is flagged as spam, but they do not stop a screener from answering first. The only way through is to answer the screener's question well.
Voicemail is a recording that waits for a message. An AI call screener is a live assistant that asks a question and decides whether to ring the owner based on the answer. Answering machine detection (AMD) is the dialer-side guess about whether a human or a machine picked up. Classic AMD was built to tell people from voicemail greetings, so it misreads screeners both ways: as a machine, so the dialer hangs up or plays a voicemail into a conversation, or as a human, so the rep is bridged to an assistant.
When an AI call screener answers, TwinsAI replies to its question naturally, the way a person would: for example, "This is John from Acme, calling about your Q3 hiring plans." The reply is set by a prompt in the TwinsAI platform, so it states your real caller and your real reason. The screener rings the prospect; when they pick up and a human is detected, the call is patched to your rep in under 450ms. It works the same on iPhone and Android screening, and it runs on every line of a parallel dial session, up to 8 at once.
TwinsAI logs each screened call to one of three outcomes: hung up, went to voicemail, human patched to the rep. That turns screening from an invisible leak into a number you can manage: what share of your list is screened, how often your reply gets the call rung through, and which openers and segments do better. It is the same data behind the 28% figure on this page.
| Dialer approach | What the screener gets | What happens to the call |
|---|---|---|
| Hang up on anything that is not a human | Silence, then a disconnect | The prospect's phone never rings |
| Drop a pre-recorded voicemail | A recording that does not answer its question | Unlikely to ring through; the message is not left in a mailbox |
| Bridge the rep as soon as anything answers | A rep saying hello to an assistant | The rep's time is spent on a machine, and the opener is wasted |
| TwinsAI: answer the screener | The caller's real name and reason, in natural speech | The screener rings the prospect; once a human is detected the call is patched to the rep in under 450ms |
iPhone and Pixel behavior checked October 2026 against Apple and Google's support documentation. The 28% figure is from TwinsAI's own call logs (500,000+ outbound calls, six months to October 2026).
It is a phone feature that answers calls from unknown numbers, asks the caller for their name and reason, and only rings the owner if the call seems worth taking, while showing them a live transcript. Apple ships it on iPhone as Ask Reason for Calling, and Google ships Call Screen on Pixel.
In TwinsAI's own logs of more than 500,000 outbound calls over the last six months, about 28% of mobile calls were answered by an AI call screener, and the share is rising as more people turn screening on.
Yes, if it answers the screener's question instead of hanging up or playing a recording. TwinsAI replies naturally with the caller's real name and reason for calling. The screener then rings the prospect, and once a human picks up, TwinsAI patches the call to the rep.
Only if the owner turns it on. In iOS 26 and later, the Screen Unknown Callers setting offers Never, Ask Reason for Calling, or Silence. Ask Reason for Calling answers unsaved numbers and asks why they are calling before the phone rings. Silence sends them to voicemail.
No. Voicemail records a message. A call screener is a live assistant that asks a question and decides whether to ring the owner based on the answer. A dialer that treats a screener like voicemail loses the call.
It removes screened calls from your connect rate unless your dialer answers the screener. Because cold calls always come from unsaved numbers, they are the calls most likely to be screened, so a dialer that cannot handle screeners undercounts your real reach on mobile lists.
Whatever your team configures in a prompt in the TwinsAI platform, typically the rep's name, the company, and the reason for the call, for example "This is John from Acme, calling about your Q3 hiring plans." It always identifies the real caller.
See how many of your mobile dials are being screened today, and how TwinsAI answers the screener and patches the prospect to your rep.