‘We have to meet the customer there’: Why Fisher & Paykel is rethinking its AI roadmap

Customers are changing how they discover brands, research products and seek help. For Fisher & Paykel, that shift is increasingly determining where and how the business invests in AI. 

“Customer behaviour is not going to change, it has changed,” Rudi Khoury, Fisher & Paykel’s chief digital officer, told MARKETING-INTERACTIVE at Salesforce's Dreamforce conference in San Francisco.

“Their research tool is a GPT of some sort and they are discovering your brand through those tools. That is what we are responding to. We have to meet the customer there because they want to use it.”

Khoury's point is that the next phase of enterprise AI is about more than automating customer service or making employees more productive. As consumers change how they research and buy, businesses need to rethink the systems, processes and information underpinning those experiences. It also raises a new challenge for marketers.

“You've got to do things like make sure your brand is discoverable, and then when you're there, that the information is accurate,” Khoury said.

Fisher & Paykel is responding by extending AI across its business, from product design and customer service to sales. But Khoury is particularly interested in what comes next: applying the technology to growth and the changing customer journey.

The appliance maker's experience reflects a broader shift among Australian and New Zealand enterprises. At Dreamforce, business leaders described similar pressures to move beyond AI experimentation and fundamentally rethink how a business operates.

Beyond agents

At Fisher & Paykel, AI has already moved well beyond the experimental stage. Khoury said there is scarcely an area of the business that is not using the technology at scale or beginning to do so.

Its most visible Agentforce deployment is in customer service, where an AI agent on the company's website is helping consumers resolve enquiries. But getting the technology in front of customers is only the beginning.

Unlike conventional software, which follows the same predetermined workflow each time, generative AI introduces a different set of operational requirements.

“I think that a lot of folk concentrate on how they are going to take cost out with AI, but actually there's also the change in how you operate,” Khoury said.

Fisher & Paykel now has employees monitoring the agent's performance, identifying gaps in its knowledge and determining what additional capabilities customers need.

That might mean improving the information available to the agent, introducing a new automation or finding a better way to handle a particular enquiry.

The agent can scale well beyond the capacity of an individual employee, but it still needs supervision.

“They still require management and leadership like a person does,” Khoury said.

It is an important distinction as businesses begin treating AI agents as part of their workforce, rather than technology that can be switched on and left alone.

Khoury said Fisher & Paykel is already approaching its agents in that way, using testing tools to simulate customer questions, compare expected responses and monitor performance over time.

The organisation also needs to watch for changes as underlying AI models are updated.

“We're monitoring that like an employee, performance managing it,” he said.

Khoury said the opportunity is not simply to make its existing service operation more efficient, but what the company learns from customer interactions to identify where the business should invest next.

Xero's approach to scale

Xero is confronting a similar challenge, albeit with a different customer base and operating model.

Nigel Piper, who works in the accounting software company's service success operation, said scale had long been a constraint on how Xero supports customers.

Traditionally, customers would search its knowledge base and raise a support case when they needed assistance. The introduction of its service agent has allowed Xero to offer a more conversational experience, bringing customer information and different support channels into one place.

Piper said the company launched the agent to its full customer base in April, giving approximately five million customers worldwide access to the service. But he stressed that the objective was not simply to automate more enquiries.

“For us, it was very much around the experience that we wanted customers to go through,” he said.

The technology also creates capacity for Xero's specialists to spend more time with customers who need deeper assistance.

Looking ahead, Piper sees an opportunity for agents to take action on customers' behalf, moving beyond answering questions to completing tasks that have traditionally required a person. That could include a password reset or resolving a banking-related issue within Xero's platform.

Piper said Xero’s ambition is to improve the overall customer experience, rather than simply reduce the number of support interactions.

“They want to be able to give you information once, they want you to be able to use that. They want channel choice. They want immediate response,” he said.

Pilot purgatory

There is another challenge confronting enterprises: turning AI experimentation into meaningful business outcomes.

Frank Fillmann, Salesforce's executive vice president and general manager for Australia and New Zealand, said conversations with CEOs and boards across Australia and New Zealand are increasingly focused on the cost of AI, the risks it introduces and the return businesses are receiving.

While technology has made it relatively easy for teams to build something interesting, getting it into production and ensuring it delivers value, is another matter entirely.

“Everyone's got a pilot. Everyone wants to vibe code something really interesting. Everyone wants to have a hackathon and build something cool and then it goes nowhere,” Fillmann said.

He borrowed a line from Kogan chief technology officer Goran Stefkovski, who had also presented at Dreamforce.

“If it's not in production, it doesn't matter. I do think that needs to be our frame, because if it's not rolled out in production, then how do you take experimentation and actually benefit from it? It doesn't graduate up. It becomes something that is stuck in pilot purgatory.”

For Khoury, however, the issue goes beyond whether a project makes it into production.

Businesses can deploy AI successfully and still fail to realise its broader potential if they focus on automating isolated tasks rather than rethinking the processes around them.

“This is not a new problem with AI,” he said. “Value in AI happens when you're solving problems cross-functionally or across end-to-end processes.”

The starting point, he argued, should be understanding the business problem and the experience the company wants to deliver before deciding which technology can help. That approach is also influencing how Fisher & Paykel invests.

Khoury said the business initially explored building its own customer-facing AI solution, but stepped back and partnered with Salesforce, including its forward-deployed engineering team.

The priority was to establish a trusted system for customer interactions without committing the company's resources to building and maintaining everything itself.

Follow the customer

At Fisher & Paykel, the next phase of its AI journey is increasingly about growth. Khoury said customer service is becoming a well-established use case, but the same technology is moving into sales, marketing and commerce.

The danger is allowing the availability of AI tools to dictate the roadmap.

Instead, he wants the company to follow changes in how customers research products, discover brands and make purchasing decisions.

“We try to follow what our customers want,” he said.

That is also why he sees AI as a business-wide transformation rather than a series of technology projects.

Customers are moving into new interfaces. Employees are changing how they work. And the information businesses make available to AI systems is becoming part of how their brands are discovered and understood.

The implications, Khoury says, are significant. Being present in the customer's journey increasingly means understanding how AI tools find, interpret and present information about a brand.

“If we don't respond we're going to get left behind. We've got to keep up, we've got to move and the technology is moving quicker than the mobile and internet era ever did.

“Our customers are changing faster than they ever have and we've learnt the lessons from the last two decades that don't sleep on digital transformation. I think it's a combination of those two. It's the internal, pressure of making sure that we're responding, and it's the external pressure that customers are legitimately changing.”