Customer Service Automation
Telstra wanted to let children and parents call Santa from its nationwide payphone network during the Christmas season, but a human-only hotline could not answer large volumes of wish-list and question calls in real time while consistently staying child-appropriate and privacy-aware.
Reported impact
Telstra turned its Santa hotline into a real-time generative AI voice service that can answer callers dynamically across its payphone network while enforcing child-safety, privacy, and conversation-isolation controls.
Telstra built a seasonal voice experience that turns public payphones into an AI-powered hotline to Santa. Inbound calls from Telstra payphones are handed into Customer Engagement Suite from Google Cloud, which fronts the interaction and lets callers speak naturally instead of navigating a fixed script. Google’s Gemini large language model generates Santa’s answers in real time using a custom persona, prompt-engineered guardrails, and child-appropriate Christmas-only response rules. Gemini’s custom voice capabilities synthesize the spoken reply from voice recordings made for the project so callers hear a branded Santa voice rather than a generic TTS output. Calls are transcribed in real time for monitoring, transcripts are deleted after 72 hours, and each conversation is isolated so the model stays “frozen” and is not trained on caller data.
Children and parents dial in through Telstra payphones, and the call is handled by Customer Engagement Suite from Google Cloud. The conversation is transcribed in real time, Gemini uses prompt-engineered rules and a custom Santa persona to generate a response, and Gemini voice capabilities speak the answer back to the caller. Transcript data is monitored for quality control, retained for up to 72 hours, then deleted.
"Santa will use Google’s Gemini large language model (LLM) and Customer Engagement Suite from Google Cloud to answer everyone’s calls."
"Santa’s voice is generated using Gemini’s custom voice capabilities, which were created from recordings we made with Santa especially for this project."
"While the calls themselves will be transcribed in real-time to make sure that the AI program is working correctly, they will be deleted after 72 hours to help avoid any privacy concerns."
"With the tech from the Google Cloud team, Santa is able to respond dynamically and in real-time – something that wasn’t possible before generative AI."
"the conversations Santa has with each caller are isolated; this means Santa’s LLM isn’t being ‘taught’ based on any of the calls made to the North Pole. It’s ‘frozen’."
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Google Cloud Sample Archived · Sep 9, 2026 AT&T AT&T needed to route customers more accurately, increase autonomous issue containment, and make virtual assistants feel more personal instead of forcing users through repetitive phone trees and stateless support… 20% more calls autonomously BigQuery · Gemini Enterprise for Customer Experience Google Cloud Sample Archived · Aug 29, 2026 Bell Canada Bell needed to move beyond rigid support decision trees and fragmented agent tooling so it could resolve more customer inquiries autonomously while still personalizing interactions across chat, voice, and human-assisted… Handled 16M+ chat interactions, with over half resolved by AI Agent Assist · BigQuery · Gemini for customer experience Google Cloud Sample Archived · Aug 5, 2026 Cox Cox needed to move beyond traditional IVR flows so customers could explain issues naturally, resolve more requests at first contact, and shift routine support into digital channels without losing context when a live… 10% improvement in customer resolution within its first week Gemini Enterprise for Customer Experience 