Phonero achieves 59% automation with Zendesk AI agents

Phonero is a Norwegian mobile communication provider that makes mobile telephony simple and more affordable for businesses and residential customers. The company offers 4G LTE mobile voice and data, SMS/MMS, international calling, roaming, and value-added services. Since 2008, its corporate vision has focused on delivering excellent customer experiences, believing that customer satisfaction is key to success.

Dialing up greater efficiencies

In 2021, Phonero launched a companywide initiative to improve the efficiency of its internal and external processes. An underlying goal was to encourage more customers to use self-service tools. The support team, consisting of about 50 agents handling chat, email, and phone inquiries, needed a way to manage growing demand.

Marius Kaafjord, who transitioned from being a support agent for large accounts to Senior AI Trainer, saw potential in automation. He proposed the idea of using AI-powered agents in the help center to help handle repetitive inquiries.

“I explained to management that AI could save us significant costs,” says Kaafjord. “If we could reduce chats by 20 percent, the AI agent would pay for itself. That became our benchmark and we finally got the green light to go ahead.”

Launching AI agents quickly

Phonero evaluated providers like Boost.ai, used by its parent company Telia, but ultimately chose Zendesk Suite for its competitive pricing and automation capabilities.

“Zendesk stood out because it was cost-effective and easy to set up,” explains Kaafjord. “Its no-code platform meant we didn’t need extensive IT resources to integrate it with our other systems.”

Phonero was able to get started quickly with its AI agent, which uses generative AI to understand a customer’s query and then formulates a response based on the information in the company’s knowledge base.

The team launched the AI agent in just six weeks with 30 to 35 intents focused on common customer queries, such as requesting invoices and ordering SIM cards. Six months into their journey, they’ve expanded to 154 intents, providing an even greater range of specific answers to customer queries, drastically increasing their automation rate.

In just a few months, Phonero’s AI agent was resolving 20 percent of customer queries, meeting the team’s benchmark.

Expanding capabilities with zero training and APIs

Over time, the company expanded the AI agent’s capabilities. For instance, the team has leveraged zero training, which has allowed them to address new customer inquiries even faster without extensive training or administrative overhead.

Haris Segetalo, a Senior AI Trainer who started in Phonero’s customer support department in 2010, has been working with zero training in a test environment and loves the technology.

“Zero training makes it much easier to implement new intents for unique or uncommon questions without needing extensive training data,” he says. “Currently we need 80 to 100 cases to accurately recognize what the customer is asking. With zero training, you just describe what the question is or what the customer needs help with, and it works right from the start.”

The team also added API functionality to verify permissions for tasks like subscription changes. Segetalo explains, “So if an administrator wants to make a subscription change through the self-service portal, the APIs verify whether that individual is authorized to do that or if additional approval from their department leader or employer is required.”

Realizing ROI in just six months

Today, Phonero’s AI agent handles roughly 50,000 chats annually, up from 17,000 chats when the automation project began. AI agents also resolve 30 percent of customer emails per month, and provide 24/7 support.

“We moved a lot of traffic from our phone and email channels over to chat,” reports Kaafjord. “In just six months, the AI agents had already resolved enough chats to cover their own costs plus the cost of us working on them.”

Phonero’s automation strategy reflects key steps found in Zendesk’s Road to +80%, a roadmap for advancing automation and integrating AI in customer support.

Road to +80% outlines how companies can quickly achieve 10 percent automation by using generative AI to create a knowledge base and launch an AI agent. For 20 percent, they can add conversation flows to control and customize responses to more sophisticated queries.

By using APIs to integrate AI agents with backend systems, companies can realize a 40 percent automation rate. And to reach 80 percent automation, they can expand AI to new use cases and channels.

AI upfront work lessens load on human agents

The AI agents also save agents time by providing customers with solutions they can try on their own before escalating issues. For example, customers who need to order SIM cards can go to the self-help portal where the AI agent provides clear, step-by-step instructions on how to order new cards directly in the app.

For queries related to troubleshooting phone connectivity, the AI agent handles the initial steps and collects necessary examples, reducing the time agents spend on those types of calls by about 10 minutes each.

“So instead of spending two hours a day on technical calls, now I spend maybe 30 minutes,” reports Segetalo. “The AI agent does most of the work upfront.”

Moving forward, Phonero continues to refine its automation strategy, integrating generative AI and APIs across chat and email to further enhance efficiency. The team is also exploring new features like multi-touch email capabilities and updating its knowledge base for better self-service experiences.