You deployed AI in your contact center. Now what? (feat. Ryan Strategic Advisory)

Episode topics

• 00:00 — Is your CX AI delivering or just running?
• 04:51 — What is driving the adoption of agent assist models?
• 06:41 — Why does tool proliferation create enterprise system confusion?
• 09:51 — How does Fuel iX™ Agent Trainer improve agent proficiency?
• 12:02 — Why are legacy quality assurance models failing in modern contact centers?
• 13:08 — What is the operational framework of the AI infinity loop
• 15:31 — How do forward development engineers bridge the operational gap?
• 17:11 — What are the foundational steps of a CX strategic assessment?
• 18:42 — Why is comprehensive change management critical for AI adoption?
• 21:47 — How should enterprise leaders audit their contact center automation investments?

Transcript

[00:00:00] Robert Zirk: You've made the investment. AI is running in your contact center, agents are using it and interactions are being assisted in real time. But when asked what it's actually delivering, do you have an answer?

[00:00:14] According to research from Ryan Strategic Advisory, commissioned by TELUS Digital and featured in the Enterprise CX AI: 2026 Global Survey, most enterprises intend to build the capabilities that would answer that question. Across every AI capability measured, planned investment significantly outpaces what's currently in place. The ambition is there. The foundations are still catching up.

[00:00:40] Erin Walker: Enterprises have moved fast on AI deployments, but the investment in making the AI perform hasn't kept pace.

[00:00:46] Robert Zirk: Deploying AI is an investment decision you make once. But building the quality infrastructure that measures and manages AI performance is an ongoing commitment you have to keep.

[00:00:58] Peter Ryan: If organizations are not investing in these tools and technologies, they're missing out. They're going to lose customer loyalty if there is a gap when it comes to the extent to which quality service is being delivered.

[00:01:10] Robert Zirk: Today on Questions for now, I'm joined by Erin Walker, global vice president of CX AI at TELUS Digital, and Peter Ryan, president and principal analyst of Ryan Strategic Advisory, as we ask: You deployed AI in your contact center. Now what?

[00:01:33] Welcome to Questions for now, a podcast from TELUS Digital where we ask today's big questions in digital customer experience. I'm Robert Zirk.

[00:01:47] To get a clear read on where enterprise CX AI actually stands, TELUS Digital commissioned a benchmarking study through Ryan Strategic Advisory, reaching 815 enterprise CX decision makers across North America, Europe and Asia Pacific. The picture that came back is one of a market that moved fast and is now figuring out what it bought.

[00:02:08] More than half of the enterprises surveyed now operate some form of human plus AI model in their contact centers. AI-assisted agent interactions — when an agent handles the conversation while AI surfaces information, suggestions or automations in real time — is the single most common model in production today, leading in six of the seven functions the survey measured.

[00:02:32] The survey also shows that CX budgets are largely holding or growing. 40% of enterprises reported increased AI investment heading into 2026 and nearly half are holding flat. That's a signal of continued commitment. But at that level of investment, organizations are looking for accountability. The question shifts from whether to deploy AI to whether what's been deployed is actually performing to the level that's expected.

[00:03:00] Peter Ryan has been tracking the CX market for two decades. His firm, Ryan Strategic Advisory, conducts research across the CX and BPO ecosystem, tracking enterprise buying behavior, vendor positioning and delivery strategy. I asked Peter where the market actually stands and his read, built on a decade of survey data, indicates that there's a lot more stability than some headlines suggest.

[00:03:26] Peter Ryan: I definitely think that we're now at a maturity curve within customer experience management. We're not seeing wild swings back and forth in terms of what organizations are looking for, perhaps in terms of outsourced services, what their investment priorities are. Those who are running customer experience are in a position where they're seeing a lot of consistency around what they're looking to do, what they're looking to achieve, some of the tasks or KPIs they have had put on them.

[00:03:54] Robert Zirk: Erin Walker helps global brands turn AI deployment into AI performance. As global vice president of CX AI at TELUS Digital, her role is at the intersection of AI strategy and actual delivery — ensuring the outcomes clients are after are the ones they realize. That ground-level perspective informs how she views the CX industry's approach to AI investments.

[00:04:17] Erin Walker: I actually see this as a customer experience and a growth play that we really need to be doubling down on in order to be able to drive those utopic scenarios and win the moments that matter most for our customers and for their customers.

[00:04:28] Robert Zirk: The focus on growth — in addition to cost — is one of the more significant shifts in how serious CX operators are now thinking about AI. It changes what gets built, what gets measured, and what counts as success. It also makes the question at the center of this episode more urgent: you deployed AI in your contact center. Now what?

[00:04:51] The dominant model in production right now is agent assist, which is AI running alongside human agents in real time during live interactions. The survey data shows it leading the market by a significant margin. When I asked Peter what's driving that particular configuration, his answer came back to what agents have been asking for for years.

[00:05:11] Peter Ryan: I can recall over the years, one of the biggest concerns agents have brought to the table, and quite frankly, one of the biggest reasons why they've tended to attrit in the past is because they say that they're tired of dealing with either a digital interaction or a voice-based interaction and having to toggle back and forth between multiple databases, between multiple windows that they have open.

[00:05:33] Using these tools that are AI-powered provide the agents with a level of accuracy and efficiency that, quite frankly, would've been unthinkable even a decade ago. But now these are very much game changers, not just in terms of developing a great agent experience, but also the experience for the consumer.

[00:05:51] Robert Zirk: With AI handling information retrieval, system navigation and real-time surfacing of the right answers, agents can focus on the customer in front of them rather than the tools behind them. That shifts what a contact center interaction actually feels like — for the agent and for the customer.

[00:06:10] Peter Ryan: I'm hearing far fewer people complaining about the extent to which they're having bad interactions. It used to be almost like a popularity killer when you would go to a party and say you work in the contact center or CX space. It's not that anymore. There's definitely an improvement because the agents have these tools at their fingertips that are helping them improve.

[00:06:29] Robert Zirk: But the enthusiasm for AI tools on its own doesn't automatically translate into coherent deployment. It's one of the most consistent patterns Erin sees across the enterprises she works to support.

[00:06:41] Erin Walker: I think a thoughtful enterprise starts with the outcomes that they're trying to drive and then works backwards from there, versus some companies that are actually just trying to start with an AI tool, plug it into their system and figure out where it can work best.

[00:06:55] Even when we first unleashed the power of AI and enabled our operations teams to develop their own co-pilots, it was exciting to see the level of adoption, but it actually created some confusion and convolution 'cause we created so many different disparate co-pilots within the front line that there was inefficiencies and a lack of understanding as to which ones to use.

[00:07:13] So we had to take a step back, recalibrate and then create a more unified implementation and delivery in order to get to those end results.

[00:07:21] Robert Zirk: That experience of too many tools and not enough architecture, is one of the most common patterns in the enterprise market right now. And when you ask enterprises how they're sourcing their AI capabilities, the answers fragment quickly.

[00:07:35] Some are working through their CCaaS provider. Others are relying on third-party point solutions. Some are combining both and a small number are building custom. The result is a market made up of many different strategies. It's a picture Peter has seen before.

[00:07:52] Peter Ryan: I'm old enough to remember, back in the late 1990s, early 2000s, the dot-com revolution, and there was a lot of confusion about what was the right business model to use, what was the right approach to take in terms of getting onto the internet to drive e-commerce, to drive more information that was going to revolutionize the economy.

[00:08:11] I think we're still in a shakeout process and different organizations have different ways of approaching. There's gonna be some trial and error. But watch this space over the course of the next few years, and in my view, you're probably gonna start to see a lot more baked in in terms of some of the best practices in trying to develop these AI-powered solutions.

[00:08:29] Robert Zirk: That consolidation is going to take time. But in the meantime, enterprises are managing investment across multiple capability categories simultaneously. An AI stack that keeps adding layers — each one solving a different problem, while none of them fully communicate with one another.

[00:08:47] Peter Ryan: I would not want to be the CTO or the COO of an organization's contact center or CX division having to put that together. I mean, it's a jigsaw puzzle that literally keeps on taking different shapes.

[00:08:59] We're still in a relatively early stage of the shift towards AI-powered tools and AI-powered solutions. What I think will happen is over the course of the coming few years, this level of complexity is probably going to become more manageable as more tools come into play that are going to be interchangeable and that are gonna work together.

[00:09:21] Imagine the days going back to the early 2000s where you literally had five or six different channels that a contact center needed to manage. Now, you're in a position where all those different elements are effectively combined into one in terms of the management, in terms of the CRM, in terms of the information that's being pulled out from an analytics standpoint.

[00:09:43] I think we'll get to that point using AI-powered tools, but again, it's going to take a little bit of time and nobody should expect overnight solutions.

[00:09:51] Robert Zirk: The jigsaw puzzle problem Peter describes is exactly what Erin sees when she walks into a new client engagement. Her answer is less about sourcing and more about sequencing. To make it concrete, she maps out what that looks like for agent proficiency, drawing on Fuel iX™ Agent Trainer, a simulation platform built on TELUS Digital's award-winning Fuel iX™ engine, that uses AI to role-play realistic customer conversations.

[00:10:17] Erin Walker: Pick the one problem that you want to solve and trace it through the entire ecosystem, because the way to actually maximize the outputs is to have integration across the different components.

[00:10:27] If you wanna look at how you can improve agent proficiency, you can deploy Agent Trainer in isolation, and yes, you'll have the ability to potentially upskill or uplevel a specific skill set. But where it actually hits the utopic state is when you orchestrate it across the entire value chain, where your recruitment team is able to have a better appreciation of what the skills of the agents are that are coming on board.

[00:10:48] So then you can create customized onboarding paths and customized training paths leveraging Agent Trainer. You can take the outputs of where there's pain points from the agents and feed that back into recruitment so they can enhance their job profiles in the future. And then you can also feed it into operations, so then the team leaders know where they should be focusing their attention in order to help coach. And then you augment that on with real-time quality checks that give real-time feedback, but then we can aggregate that to then augment even better scenarios that the agents can then trial day in and day out as part of their agent training ecosystem.

[00:11:19] So really where you hit that utopic state is if you define what the problem is and then you look at how the different point solutions can work together in order to get the desired outcome.

[00:11:28] Robert Zirk: That's the system working as designed. The challenge is that many organizations built their AI stack faster than their strategy could keep up with it. Peter's advice to leaders at those organizations is direct.

[00:11:41] Peter Ryan: What's most important is to understand what the company's priorities are from a customer experience standpoint as well as from a broad standpoint in how the CX department can help achieve the broader commercial goals, number one.

[00:11:54] And then, number two: determine what needs to be invested in within the CX department in order to make certain that those goals are going to be achieved.

[00:12:02] Robert Zirk: 68% of enterprises running AI-assisted customer experience have no AI-powered quality assurance in place to keep pace with it. Peter argues that QA in most contact centers is still built for a different era despite the advancements in how the interactions themselves are handled.

[00:12:20] Peter Ryan: Let's think a little bit about the days when you used to have quality assurance that was performed by a department and the department had individuals who would spend time listening to telephone calls. Maybe they'd get two, three, four done in one day if they were lucky. They would make notes, and then they would use a randomized sample of those calls and identify who was performing well, who wasn't performing well and they would look to try and plug the holes.

[00:12:46] Now, with quality assurance tools that are powered by artificial intelligence, you can literally listen to 100% of the calls in a very short amount of time and be able to identify any red flags as well as individuals that are doing exceptionally well, so that you can have a real sense in near real time about how well the entire department's performing.

[00:13:08] Robert Zirk: That coverage is what makes a broader framework possible. Erin calls it the infinity loop.

[00:13:14] Erin Walker: The infinity loop is really about recognizing that all of these things are not separate programs and that they're a cycle and they need to talk to each other. So when you hire, you're setting the foundation of what agents can do.

[00:13:24] So how do you better leverage quality insights to feed into the job profile so that we can use AI in order to be able to do some evaluations and find the right people for the right roles so they can be more successful from day one? Then how do we feed the outputs of that the training curriculum, so we can have very customized training paths for our agents, so that we can train them on the skills that they need the most amount of support on.

[00:13:46] And then how do we flag those opportunities as we move them into production, and evaluate the agents in real time to have immediate coaching and feedback, which we all know is a lot easier to be able to absorb and iterate on than having feedback once a month on a small sub-subset of the calls that you've taken.

[00:14:03] On top of that, if you can integrate it into the workforce management team, who can automatically identify who low performers are and pull them out of the load so they can do some customized training within our Agent Trainer environment, then you've got the utopic ecosystem talking to each other across that broader infinity loop.

[00:14:19] Robert Zirk: The quality problem isn't only an operational one. For enterprises that haven't addressed it, Peter sees a clear commercial exposure.

[00:14:27] Peter Ryan: If organizations are not investing in these tools and technologies, they're missing out because they're going to lose customer loyalty if there is a gap when it comes to the extent to which quality service is being delivered.

[00:14:41] Robert Zirk: So how do CX leaders begin to build the proper infrastructure to measure and manage their AI? For Peter, the first principle is that you can't shortcut the work of making sure technology actually performs before it goes live.

[00:14:55] Peter Ryan: AI, as a technology, is an enabler of existing tools. Just because something has a specific AI component doesn't mean it's going to work. It needs to be stress-tested. It needs to take into account nuances, perhaps of a particular industry or of a particular location.

[00:15:12] The reality is that you can't put something out there that's not going to work. You're just going to erode your consumer's loyalty and we've all had these experiences. Where I'm coming from is now it's probably not just more important than ever, but in many cases more straightforward than ever in terms of making these optimizations and making sure that everything's rock solid.

[00:15:31] Robert Zirk: That stress-testing discipline is about catching failure before it scales. Erin's approach at TELUS Digital takes it a step further, embedding the people who build the technology directly into the operations that use it. These team members are called forward-deployed engineers, and this approach directly addresses the gap between what gets built separately and what works on the floor.

[00:15:55] Erin Walker: Instead of building something in isolation and then handing it over to the delivery arm or to the operations arms, our engineers are embedded directly into the operation. So they're sitting side by side the agents, side by side your real-time analysts, your schedulers, your forecasters, your quality team members, watching the work happen in real time, and as I referenced earlier on, really appreciating what's working well, what's not, how are the individuals adapting the tools, how are they bypassing the tools and then iterating and evolving the technology in real time in order to really be able to maximize the benefit that we're seeing.

[00:16:28] It really solves the classic gap between what the vendor builds and what operations actually needs.

[00:16:33] Robert Zirk: Erin described one recent example that illustrates forward-deployed engineering in action: a notification tool behaving perfectly in testing that created a real problem once it went to the floor.

[00:16:45] Erin Walker: We created one tool that was able to give real-time notification in regards to interval compliance of the agents. But what we realized: it was pinging too high of a frequency, and it was pinging on a desktop when the team leader was actually walking around the floor dealing with fires on a regular basis. So being able to quickly identify that gap and look at how do we modify the volume of pings and actually move it to a mobile versus a hard desktop environment allows us to net much stronger outcomes from that perspective.

[00:17:11] Robert Zirk: That kind of embedded iteration is one of the tools in the TELUS Digital toolkit. Another is what Erin describes as a CX Strategic Assessment, a structured diagnostic designed to surface where the real friction is in an operation before anything gets built.

[00:17:28] Erin Walker: We come in, we take a full look of what's going on within your CX environment. So we're looking at what are your priorities, what are your current workflows, where do you have the highest volume of manual work or inefficiencies? What is the data telling you in regards to trending or opportunities? Where have you deployed AI? Where haven't you? And we aggregate all this information to put forth a series of recommendations as to how you can maximize stronger outcomes for both your business as well as your customer, looking all the way through your digital footprint to your workflows through to how you manage within the call center environment day in and day out.

[00:17:59] And there's a lot of pretty cool ideas put forth through that both solving specific problems, as we talked about, or a broader innovation roadmap as to the different elements that you can put in place over the course of a three to five-year horizon.

[00:18:11] We had one customer where there was a point of friction in regards to their quality workflow. So poor case prioritization, constant switching of work, unclear escalations, so the right people were not doing the right work and it was causing a lot of friction in the agent experience as well as the customer experience.

[00:18:28] And so we partnered with this organization to build a dashboard with an intelligent notification widget, so AI-driven priority lists that we're able to move in an agile way, making sure that the work got to the right agents at the right time to drive the strongest outcomes.

[00:18:42] Robert Zirk: Erin highlighted an element of AI delivery that often gets overlooked on checklists: getting people to trust, use, and champion it.

[00:18:51] Erin Walker: I think everyone underestimates change management and how the agent needs to trust AI, how they have to buy into AI, how it has to work for them successfully, hopefully the first, if not the second time with the FDE right beside them. But that change management component of it is exceptionally, exceptionally important.

[00:19:09] We spend a lot of time not only with our engineers on the floor customizing, adapting it, ensuring that it's fully integrated into the processes to drive the strongest outcomes. But we also have AI influencers, recognition programs, super users, call-a-friend tools, usage boards, T-shirts and badges, and all sorts of fun stuff to get everyone excited about what we're trying to do, which is a really critical component of the overarching change management.

[00:19:34] Robert Zirk: The AI-powered infinity loop only works if the people inside it believe it's trying to help them and if the data it generates is being captured, measured and fed back into how agents are trained, coached, and supported.

[00:19:49] Looking ahead, I asked both Peter and Erin where they expect the market to move over the next 12 months. Peter noted that, despite the current fragmentation, he expects fewer competing approaches and better defined investment priorities within the next year. What he's less certain about is whether CX will actually be treated as a strategic priority by the organizations that need to make that shift.

[00:20:13] Peter Ryan: There's still a lot of confusion in the market right now, and where I'm coming from, and I speak at a lot of conferences, one of the things that you get a real sense of is that there's no coalescing around what the right approach to taking on AI is going to be.

[00:20:29] I think we're gonna start seeing a much better framework about how organizations want to invest. I think we'll get a better idea about what the priorities are gonna be, where the investment money needs to go and to the extent of which CX is going to be a major priority for different organizations.

[00:20:47] Right now we hear a lot of rhetoric, and if you talk to CEOs, they'll all say, "Yes, customer experience is a priority for us." I'm not so sure it is. I think it's gonna be much more of a priority in twelve months. As we talked about, consumers are more willing to attrit and to shift than they ever have been. Any CEO is going to understand that a major component around that is investing in the right tools, the right processes and in the people. All of these things are important from the standpoint of where I think the market's gonna be going twelve months from now and in the longer term.

[00:21:17] Robert Zirk: Within the next year, Erin expects more integration across point solutions, higher adoption rates and most urgently, more rigorous optimization work for the AI that's already been deployed. It's a challenge she sees across the industry.

[00:21:33] Erin Walker: A lot of times we're investing in AI, but we still haven't justified fully the investment and we're not seeing the full benefit yet. So, collectively, I'm hoping organizations are able to better synthesize that down to be able to drive the ROIs based on the investments being made.

[00:21:47] Robert Zirk: I asked Erin to put that into a question CX leaders should be asking themselves.

[00:21:52] Erin Walker: Do I know with confidence which of my AI investments is actually moving outcomes and how do I orchestrate them together effectively in order to get the desired end goal? it comes back to the earlier theme that we were talking about. What is the problem that you are trying to solve with AI? Really appreciating what that problem is and then driving the right level of tracking against being able to solve that problem.

[00:22:14] So it's not just about, again, usage or the number of deployments, but it's "Are you actually driving the business outcomes, the customer outcomes that matter most to your business and your customer by the power of AI?"

[00:22:26] Robert Zirk: Peter offered similar advice for CX leaders, acknowledging the pressure they face to deploy AI whether they're ready or not.

[00:22:34] Peter Ryan: You know, I hear this all the time. A customer experience leader goes into a boardroom and one of the first questions they get: "What are we doing with AI?"

[00:22:41] It's not a case of just deploying AI for the sake of it. It's knowing what you want to do with it. Where are the areas of customer experience within your organization that you can reasonably take on this technology that's gonna make a big positive difference? Figure out exactly what the best solutions are in the market and partner accordingly.

[00:22:57] Robert Zirk: And Erin views getting this right as, ultimately, a growth priority for brands.

[00:23:02] Erin Walker: I'm really excited and energized about the impact that we can make — not only from a cost side, but more importantly, from a customer experience and from a growth trajectory. So I think this is a really exciting time to be part of the CX industry on a global basis and definitely thrilled and humbled in regards to the impact that collectively we're gonna make in the coming months and the coming years.

[00:23:28] Robert Zirk: Thank you so much to Peter Ryan and Erin Walker for joining me and sharing their insights today. And thank you for listening to Questions for now — a TELUS Digital podcast.

[00:23:39] For more insights on the latest enterprise contact center AI trends, you can find the full Enterprise CX AI: 2026 Global Survey Report, featuring research from Ryan Strategic Advisory that was commissioned by TELUS Digital, at telusdigital.com. We'll include a link in the show notes as well. And if you found this episode useful, be sure to follow Questions for now wherever you listen to podcasts.

[00:24:05] I'm Robert Zirk, and until next time, that's all... for now.

[00:24:09] ​