"AT&T is testing an agentic artificial intelligence (AI)-powered digital receptionist that the company says has the potential to significantly reduce the headaches caused by phone spam and fraud. AT&T says that the system relies on multiple large language models (LLMs) and its long history dealing with fraud and other malicious types of calls. An advantage of the AT&T AI receptionist is that it is built into the network and does not rely on the user's phone or use its resources."
"As the name implies, the service - which is rolling out to "select customers" this year - has a conversation with callers to determine whether they are legitimate or spam. The AT&T AI receptionist will ask normal questions, such as "Who may I say is calling?" and "What is this in regards to?" The call is let through if the responses fulfill the requirements set by the user."
"The system can complete rudimentary tasks such as taking a message or accepting a delivery window on its own. The agent will either take a message or disconnect if the caller refuses to identify themselves, is a wrong number, or fails to meet the user's criteria. The user can watch a live transcript and pick up the call at any time. They can assess a text summary and determine whether or not to call back."
AT&T is testing an agentic AI-powered digital receptionist aimed at reducing phone spam and fraud. The system leverages multiple large language models and AT&T's experience with fraudulent and malicious calls. The receptionist is integrated into the network, so it does not consume the user's phone resources. The service converses with callers, asks verification questions, and allows calls through only when responses meet user-set criteria. The agent can take messages, accept delivery windows, and disconnect unwanted calls. Users can view live transcripts, take over calls, or review text summaries and maintain a do-not-screen list for trusted contacts.
Read at Telecompetitor
Unable to calculate read time
Collection
[
|
...
]