TELUS Digital Cuts Contact-Center Onboarding Time 20% With AI Voice Simulations
TELUS Digital’s vice president of product, Mitch Lieberman, presents the company’s Agent Trainer as a response to a high-volume contact-center onboarding problem: 70,000 associates, 20,000 to 30,000 hires a year, and industry churn of 30% to 50%. Built on ElevenAgents, the voice and chat simulation platform is intended to get new agents ready for customer interactions faster, with TELUS Digital reporting a 20% reduction in time to proficiency, more than 50,000 completed simulations, and early signs of lower churn.
The operating problem is speed to proficiency across 20,000 to 30,000 hires a year
Mitch Lieberman, identified on screen as TELUS Digital’s vice president of product, frames the contact-center training problem as a throughput problem. TELUS Digital has 70,000 contact-center associates, and Lieberman says annualized agent churn in the industry runs 30% to 50%. For TELUS Digital, that translates into hiring roughly 20,000 to 30,000 people a year.
In that context, reducing onboarding time is tied directly to operating scale. Lieberman describes a 20% reduction in the time it takes to make agents proficient, or ready to take calls, as “a 20% operational savings times 20 to 30,000 per year.”
The training target is what Lieberman calls “speed to proficiency”: getting new associates to the point where they can provide customers with an equivalent satisfactory response, measured through customer satisfaction or Net Promoter Score. Traditional classroom sessions, shadowing, and occasional roleplay did not cover enough of the real-world scenarios agents face. Lieberman’s spoken emphasis is narrower and operational: TELUS Digital needed to bring new, non-tenured agents up to speed quickly.
Chat simulations were understood; phone-call simulations exposed the gap
Lieberman says the initial proof of concept with a customer, conducted early in the fall, confirmed demand for training around speed to proficiency. For chat-based work, the pattern was already familiar: an agent converses with a chatbot, and the system evaluates whether the agent answered correctly.
The harder problem came from voice work. TELUS Digital heard from customers that roughly half of their agents answer phones. If chat could be simulated for training, the company needed a way to simulate phone calls with comparable usefulness.
That requirement changed the technical bar. The system had to simulate real calls, not just exchange text. Lieberman emphasizes latency: the delay between one person speaking and the other person responding. To feel like a phone call, the response time had to sit inside a normal conversational delay. TELUS Digital also needed high-quality voices and voice selection.
ElevenLabs provided the end-to-end capability TELUS Digital needed, while also leaving room to “take it apart and stitch together with other solutions” if necessary. TELUS Digital did not ultimately need to do that, but the option mattered because it distinguished the platform from more closed systems.
The platform choice came down to latency, control, cost, and build speed
Mitch Lieberman says ElevenLabs ranked in the “top quartile” and “probably top 10%” among the evaluated platforms for latency. But latency alone was not the full differentiator. Other platforms came close on latency, he says, but were more expensive and operated as black boxes.
The advantage was a balance between packaged capability and developer control. ElevenLabs’ conversational AI tooling, now part of its agent AI platform, let TELUS Digital prototype directly and move to market quickly. TELUS Digital went from selection to live deployment in five months.
ElevenLabs has the right balance of tooling that allowed us to use conversational AI, which has now become agent, your agent AI platform, so that we could build directly and we can prototype and go go to market exceptionally fast.
The same theme appears in Lieberman’s description of product prototyping. One differentiator was that “anybody” could enter the platform, prototype, and understand “the art of the possible.” As a product leader, he was able to build prototypes himself and hand them to engineering.
That mattered because TELUS Digital was not merely buying a finished training product. It was building an agent-training platform on top of ElevenLabs. Lieberman presents the platform as sitting between two extremes: a black box, which some buyers may want but TELUS Digital found limiting, and a complete toolkit that requires teams to write everything in code.
The system has moved from deployment into repeated evaluated simulations
The training platform is described as Agent Trainer on ElevenAgents: dynamic voice and chat simulations for scenarios ranging from routine inquiries to more complex complaints. Lieberman, speaking about ElevenLabs’ agent AI platform, says TELUS Digital ran 35,000 customer simulations with end evaluations over the prior six months. The broader total is more than 50,000 completed customer simulations.
| Metric | Figure | Context |
|---|---|---|
| Contact-center associates | 70,000 | TELUS Digital workforce size cited by Lieberman |
| Annual hiring volume | 20,000–30,000 | New hires per year cited by Lieberman |
| Recent simulations | 35,000 | Customer simulations with end evaluations over the prior six months, according to Lieberman |
| Overall simulations | 50,000+ | Completed customer simulations cited in the source description |
| Onboarding improvement | 20% | Less time for new agents to be ready, according to Lieberman |
Mitch Lieberman defines the result as operational efficiency and speed to proficiency: training agents in less time so they reach the same level of satisfactory customer response. The benchmark is whether new agents can be ready sooner while still delivering acceptable outcomes for the customers of TELUS Digital’s customers.
TELUS Digital has also started to see a reduction in churn. Lieberman connects that early sign to agent confidence and comfort: agents who have practiced with simulations feel more ready when they reach live customer interactions.
AI practice changed the social pressure of training
One unexpected benefit, Mitch Lieberman says, came from onboarding feedback. TELUS Digital learned through verbatims — direct feedback from participants — that some agents preferred working with an AI over a human-guided coach.
The reason was not that the AI was more expert. It was that the AI separated practice from judgment. When agents worked with a person, Lieberman says, they felt they were being trained and judged at the same time. With an AI system, they could separate those experiences.
That distinction matters for the experience of training with simulations. The platform is not only simulating customer conversations; it is creating an environment where agents can repeat scenarios, receive evaluations, and build comfort before taking live calls. Lieberman links that greater confidence and comfort to the early signs of reduced churn TELUS Digital is seeing.
The next extension is an AI voice coach after evaluation
Mitch Lieberman says TELUS Digital plans to branch the platform in two directions, with a focus on coaching. After an evaluation, an agent will be able to request help from an AI voice coach. That coach will review the evaluation and give tips on what the agent should have said.
TELUS Digital already presents feedback in text. The next step is multimodal coaching: adding the ability to have a voice discussion with an AI coach after the evaluation is complete. The proposed flow keeps the simulation, evaluation, and coaching loop within the same training environment.
Lieberman’s endorsement of ElevenLabs rests on platform role as much as voice quality. He cites scalability, security, API access, and the ability to build on top of the system. Voice quality, low latency, and voice selection were priority criteria, but so was the ability to move quickly without surrendering control to a black box or taking on the full burden of building every capability from scratch.