The problem
Customer care agents at a large carrier spend much of each call searching for answers: plans, devices, billing rules and troubleshooting steps spread across a large knowledge base. Time spent searching lengthens calls and leaves little room for anything else, including the sales conversations that a service contact can open. Verizon’s goal was to give agents answers on screen, shorten calls and free them to sell, according to Reuters (Reuters via AOL).
What they did
Verizon built an internal assistant, which Google calls the “Personal Research Assistant”, by feeding nearly 15,000 internal documents into a version of Google’s Gemini model, said Sampath Sowmyanarayan, chief executive of Verizon’s consumer group (Reuters via AOL). Deployment started in July 2024 and reached full scale in January 2025. He said Verizon chose Google over competitors for its ability to deploy AI to tens of thousands of people.
According to Google Cloud’s announcement, the assistant runs on Vertex AI, Gemini models and Google’s Agent Assist Panel (Google Cloud). It suggests questions a customer is likely to ask and answers questions agents type in, so they no longer search the knowledge base by hand. A second agent, “Problem Solver”, guides reps through advanced troubleshooting, which Google says helps new reps most. Verizon also runs Gemini-powered virtual agents for customers by phone, chat and the My Verizon app.
The workforce choice is the center of the case. Sowmyanarayan described it as “reskilling in real time from customer care agents to selling agents”, and said there are “much easier ways” to reduce costs than replacing staff, contrasting Verizon with companies such as Klarna that used AI to replace customer service workers.
By mid-2026 the emphasis had shifted. Chief executive Dan Schulman told Bloomberg TV that Verizon had spent the previous three months experimenting with AI agents that replace some customer service reps, and that AI disruption “can be a large percentage of customer service” (CX Today). He said routine requests such as forgotten passwords and billing amounts suit AI agents, while complex issues need humans and AI working together.
Results
Sales through Verizon’s 28,000-person service team are up nearly 40% since the deployment, according to Sowmyanarayan as reported by Reuters (Reuters via AOL). Reuters also reports that the assistant cut down on call times, without a figure. Verizon has not published the baseline, the period or how much of the increase it attributes to AI rather than pricing, promotions or other changes.
Google Cloud’s announcement says the assistant is deployed across 28,000 of Verizon’s customer care reps and retail stores and that the collaboration led to a 95% comprehensive answerability rate for customer inquiries (Google Cloud). That figure is vendor-reported and the release does not define the measure.
For the newer AI agents, Schulman said their customer satisfaction rate is 1,280 basis points better than before, as reported by CX Today (CX Today). He did not say what the rate is measured on or what it was compared with.
What didn’t go to plan
No public source reports a reversal, but the story changed. In 2025, Verizon presented AI as a way to keep agents and turn them into sellers; by 2026 its chief executive was describing AI agents that replace some reps and warning that a large percentage of service work could be disrupted. Leaders who adopt the first framing should plan for the second.
The evidence is also thin. Every number comes from an executive interview or the vendor’s press release, none from Verizon’s financial reports, and none has a published method. A sales rise in a service channel can also signal a risk: agents pushed to sell during service calls can frustrate customers who called for help, and the sources do not report complaint or churn data for the period.
Lessons
- Use AI to change what agents do, not only how fast. Verizon framed its assistant as moving care agents toward selling, and reports nearly 40% more sales through the team.
- Ground the assistant in your own knowledge. Verizon fed nearly 15,000 internal documents into the model so answers reflect its own policies.
- Choose a partner that can deploy at your scale. Verizon said it picked Google for its ability to roll out to tens of thousands of people.
- Expect the workforce story to evolve. Within about a year, Verizon moved from reskilling agents to testing AI agents that replace some reps.
- Publish the baseline with the headline. Without a period, baseline or method, a “nearly 40%” sales rise is hard to credit to AI.
Adapt it in your organization
Is this for you?
- Your service agents spend significant call time searching knowledge bases.
- Service contacts are a realistic moment to offer upgrades or new products.
- Your knowledge base is current enough to ground an assistant’s answers.
- You can attribute sales to the service channel and compare against a control group.
- You track complaints and churn closely enough to see if selling hurts service.
A 90-day version
Pilot: give an AI answer assistant to one service team, with a matched team working as usual, and track handle time, first-contact resolution, sales per contact and complaints for both. Owner: head of customer care. Success measure: sales per contact up at least 10% with no rise in complaints or repeat contacts.
Metric to track
Sales per service contact alongside complaint and repeat-contact rates, against a control team. Verizon reports nearly 40% more sales through its service team; for a case where service quality slipped after an AI push, see the Klarna case.
Failure modes to pre-empt
- Crediting all sales growth to AI without a control group or baseline.
- Turning service calls into sales calls that customers resent.
- Grounding the assistant in outdated documents that give wrong answers.
- Promising staff the tool will not replace them, then changing course without a plan.
Questions to take to your leadership team
- Is our goal to make agents faster, to make them sell, or to need fewer of them, and have we said so?
- How would we know if selling on service calls is hurting satisfaction or churn?
- What baseline will we publish alongside any sales figure?
Executive FAQ
How much did Verizon’s AI assistant increase sales?
Verizon’s consumer chief told Reuters that sales through its 28,000-person service team rose nearly 40% after deployment, a figure Verizon has not published in its reports.
Is Verizon replacing customer service staff with AI?
In 2026 its chief executive said Verizon is testing AI agents that replace some service reps, while complex issues still involve humans.
About the curator & AI disclosure
Curated by Luciano Ayres
Luciano Ayres is Director of Artificial Intelligence at ArcTouch, with more than 20 years of experience in software and technology leadership and specialized studies in deep learning. Connect with him on LinkedIn.
AI-generated briefing. This case was produced through an automated AI research and writing workflow curated by Luciano Ayres. Sources are linked for independent verification; summaries and analysis may contain errors and do not replace professional due diligence.