What you need to know

• AI won’t replace agents – it’ll replace admin as per Catapult, Sealy, and Jaycar, who are using Zendesk’s AI tools to eliminate repetitive tasks, improve support speed, and let human teams focus on empathy, precision, and complex problem-solving.
• Catapult has cut service handling times by 14 per cent. The elite sports tech firm implemented Intelligent Triage, reducing full resolution times from four hours to 2.7 hours while boosting agent efficiency and maintaining high-touch VIP service.
• Sealy, meanwhile, has revamped its customer service model to support direct consumer interactions, slashing average response times from days to 90 minutes and lifting agent productivity by up to 30 per cent.
• Jaycar’s fridge-named bot solves 27 solves of tickets solo. Jaycar also cut response times from four days to 18 hours, driving major lifts in CSAT and NPS scores, especially for its Brass Monkey brand.
• AI adoption is driven by workflow gains, not headcount cuts at all three brands, which stress that automation is enhancing—not replacing—human agents by improving collaboration, streamlining workflows, and delivering faster, more consistent support.
• Catapult joined its vendor's Early Access Program to co-develop features, test in sandbox environments, and ensure AI models met the complex needs of elite-level customers without compromising service quality.

AI isn’t coming for your job – it’s coming for your paperwork. At least, that’s the verdict from Catapult, Sealy and Jaycar, which have all leaned into Zendesk’s AI to sharpen service performance without sidelining human agents. From triaging support tickets to automating mind-numbing admin, the bots are clearing the decks so frontline teams can focus on what matters: Empathy, speed and solving real problems.

According to Alex Finster-Rowen, Catapult’s Global Projects Director, customers can now reach out via SMS, a web form, or email. “No matter how they contact support, agents can collaborate seamlessly. They have access to a unified dashboard, so regardless of the channel the issue comes through, it's all in one place. Agents can easily hand off the case to a specialist team without the customer needing to repeat themselves.”

Catapult’s business involves operating where sports science crashes into data analytics, and it provides tech that fine-tunes performance, flags injury risk before it happens, and puts hard numbers around return-to-play decisions.

“We’ve all experienced the frustration of being passed from one team to another and having to retell the same story each time. With omnichannel support, that doesn’t happen—the customer never has to repeat their issue, which creates a much smoother experience,” says Finster-Rowen.

“From the customer’s perspective, if they start with a chat, they can switch to email without any disruption. The agent simply continues handling the same ticket. It's a great example of team collaboration."

Since implementing the current customer service platform almost eight years ago, the company has seen a significant drop in first response and full resolution times.

“A few years ago, our first response time went down to 12 minutes. And then the full resolution time dropped to four hours, even for the most complex issues.”

The platform also enables Catapult to support different types of customers. “VIP queries, for example, require help centre content that’s completely different from what amateur teams or individual players need. So, we created multi-brand knowledge bases.”

Per Finster-Rowen: “We run nine separate help centres, each catering to different customers and products. The content is organised in a brand-specific way, so when a customer is looking for something, they can find the relevant information quickly.”

All of this was before Catapult dived into AI.

“Now we’ve started the journey with AI. We actually began looking into AI components last year. When you think about it, it's been a very short journey—especially in comparison to how long we’ve been using [the platform], which is eight years. That really shows how innovative the technology is,” she says.

AI has helped Catapult develop and innovate its products, “And we’re also constantly improving the services that support them.”

Catapult is part of its vendor's Early Access Program (EAP), which allows it to test different features and products before they’re released to the public. 

“That’s allowed us to work directly with Zendesk product managers, provide feedback, and help fine-tune the products before they go live. It’s been super beneficial—especially when it comes to AI," Finster-Rowen says. “So in 2023, our goal wasn’t just to dip our toes into AI, but to really go all in. The main objective was to seamlessly integrate AI into our team’s workflows."

The challenge was to do this without risking the relationships Catapult had developed with its customers. Execution was critical with Catapult creating a clear roadmap, setting milestones, and spending about six months testing various features in the sandbox before moving forward with a customised deployment.

“We essentially cherry-picked the features we were confident would benefit our customers, and there were some really clever ones in there. The first major component was Intelligent Triage,” she says.

Finster-Rowen offers the example of an elite sports coach submitting a query. “With Intelligent Triage, before the ticket even reaches an agent, the system reviews the context and adds useful information to the ticket. This way, agents know what type of issue they’re dealing with, the customer’s sentiment, which is key for fine-tuning, and how critical the issue is. That allows for more accurate prioritisation, which has massively improved both ticket routing and handling times.”

The result was faster service and reduced manual workload.

“Before we had Intelligent Triage, we had six support agents manually triaging a large volume of tickets. After implementation, those agents were able to shift their focus to more complex issues, which significantly improved our efficiency.”

The metrics tell the story: Average handling time is now 24 minutes—a 14 per cent improvement. Full resolution time, which used to be four hours, has dropped to 2.7 hours.

“Looking at just one AI component, we saw a 21 per cent improvement, which really shows what’s possible. The key was deploying it in the right way, so we didn’t disrupt the nuanced relationships we’ve built with our different customer groups.”

Finster-Rowen acknowledges there were significant challenges. The first was deploying AI models that could understand and respond to highly complex customer queries, especially from its elite customers, without losing accuracy or context, and without impacting those relationships.

“AI is powerful, but it’s only as effective as the boundaries you define and the context you provide. For us, that context is quite complex, so we focused on fine-tuning the AI models to meet the expectations of our elite support customers. Their standard is: ‘I want to speak to a human, I need the answer now, and it must be precise. With AI, we had to design workflows that enabled seamless, instant escalation to human agents, while still using AI bots as part of that world-class VIP experience.”

Another challenge—one that many industries face—was the transition to automated systems, especially AI agents and bots. At first, there was some hesitation from the broader support and customer success teams, she admits. The key was showing them the value.

“We demonstrated how automation could save time, reduce repetitive tasks, and free them up to focus on real, complex issues—delivering more value to our customers. And they bought into it—they love it now.”

Today, Finster-Rowen says, the implementation and rollout of new features are seamless. “The tools truly empower our support teams, and their jobs have become much easier as a result.”

Sleep easier

For mattress brand Sealy, a change in its business model provided the impetus to update its customer service platform.

According to Michael Chen, Sealy’s head of digital and CX, “I think the way we deal with consumers is very different. With retailers, we already have relationships in place—they come in quite warm. They usually know the background. But with consumers, we often only get one shot.

“A lot of the time, when consumers contact us, it’s their very first time hearing about Sealy. That’s because we sell through our retail network. Everything goes through retail—so the salesperson and the store itself are really important to us. They represent the brand. It’s important consumers feel like they made the right choice by choosing us. So it’s definitely a mindset shift.”

The way the company views customer care is also changing, he says. “Traditionally, it’s been seen as a department that fulfils legal requirements—making sure we don't get into trouble, handling claims. But to me, the culture is shifting. Now, we see customer care—or CX—as a competitive advantage.”

Whether it’s a retailer or a consumer reaching out, whether it’s a compliment, complaint, or inquiry, Chen says Sealy’s goal is for customers to walk away thinking, “Wow, this is a fantastic organisation I can trust.”

That approach also helps build trust with retailers.

“We’ll say, ‘Look, we’ve handled this many complaints, and consumers are walking away with 85 per cent satisfaction.’ That helps both them and us—it lets them focus on selling, knowing we’ve got their back if anything comes up," Chen says.

Sealy has gone long on Zendesk, having bought the full suite, according to Chen. “We have the Agent Workspace, which has been fantastic—really helping us with workflows. We use the Explorer product to report on fault rates, which is used by our manufacturing R&D. We also use Guide as our internal knowledge base, so agents can check specs, product information, how to troubleshoot certain issues and our internal processes.”

And like at Catapult, the platform’s AI is getting a big workout.

“They make our agents more efficient by reducing the time spent on repetitive tasks, so they can focus more on building that human-to-human connection. For example, during core transcripts or when filling out standard information fields, agents no longer have to manually type notes. The writing tools help us draft better responses, too, so agents don’t have to spend time on things like spell-checking.”

The impact has been significant, Chen says. “We measured agent productivity and saw an increase of about 20 to 30 per cent. They're able to handle more tickets with the same number—or even fewer—agents, and they're feeling less stressed.”

On the consumer side, satisfaction ratings have also risen.

“One of the biggest improvements has been in response time. It used to take a couple of days. During the peak of COVID, when everyone in CX was struggling—people were at home with nothing to do but complain—our standard response time blew out to a week, sometimes even two. Now, it’s down to 90 minutes.”

Responsiveness improvements

For Jaycar, the infusion of AI into customer service has seen huge improvements in response times.

“It used to take four days to respond to a customer. Now it’s 18 hours—and in some cases, the AI doesn’t even need a human,” Neil Deo, Customer Experience Optimisation Lead, Jaycar tells Mi3.

That shift, driven by a fresh customer service stack anchored in Zendesk and powered by automation and bots, is already moving key CX dials: NPS and CSAT are up, customer confusion is down, and a bot named after a fridge brand is solving more than a quarter of all tickets on its own. This is what happens when a high-churn, multi-brand support operation gets serious about transparency, macros, and machine learning.

According to Deo, “The change has been massive and that’s mainly due to that transparency, with the team being able to see each other’s tickets. We're not chewing through customers. Back in the day, as I said, the customer would reach out to us four times across four different platforms… and get four different answers.

Tackling that particular service bugbear has led to improved CSAT and NPS scores, as measured by an external provider called Resonate.

“We’re getting more insight within our store environment with our customer service. The notable change I’ve found is the response times. Customers are much better off than they were. We’ve got data from our first month of [the new platform] compared to now. Back then, the response time was about four days and right now it’s 18 hours. Could change but just that in itself is a massive improvement.

“We were never reporting on this stuff before. And then, just seeing how bad we were, where we could improve, and streamline.”

Jaycar has eight brands it serves in the network. "For one of them, Brass Monkey (Brass Monkey fridges), the AI bot auto-solves 27 per cent of interactions. It’s 10 per cent across the board for tools, 12 per cent for friends, but Brass Monkey—27 per cent. That’s just amazing,” comments Deo.

Looking ahead, Jaycar wants to further automate order inquiries. "We’ve made a lot of progress with product inquiries, but “Where’s my order?” the returns process, and getting copies of invoices—we’re working on making all of that more seamless.”

Deo says customers can currently log into their accounts and get that information, and the bot is good enough to direct them there. "But we want to be able to serve a tracking number, or serve a downloadable invoice, within a chat session," he adds.

Next big things

According to Zendesk's ANZ regional vice-president, Kellie Hackney, AI fuelled improvements in customer service are set to continue. "We'll see a lot more emerging around autonomous service. I think we'll also see further development in areas like agent-assisted AI—where, when a customer service agent gets involved, the intelligence supporting them enables more proactive responses. I believe that will be a major driver."

Another area likely to see growth is voice and voice AI. "Voice remains one of the most powerful and personal ways businesses can connect with their customers. People still want to communicate verbally with businesses, depending on the situation. So it makes sense that the better an AI agent can understand and respond to voice, the better the overall experience will be," Hackney adds.