AI supports Ropo’s customer service - resulting in higher quality and greater efficiency
Ropo strongly believes in the potential of automation technologies in service delivery. When the company began digitalizing its customer service, it identified specific parts of the process that could be automated using AI. Epical delivered a solution that saves both time and money, while also improving service quality.
Ropo wanted to utilize AI in a way that, beyond offering operational benefits, it would also enable scalability. This would make the investment easier to justify by demonstrating a clear payment period. As an international company, it was also essential that the solution could be deployed across Ropo’s countries of operation.
We started by examining our customer service process and identified two phases where automation could save time and deliver substantial benefits. The first involved classifying incoming calls, and the second was summarizing their content.
Ropo’s customer service team handles between 600,000 and 700,000 calls annually. Each call is categorized based on its topic, and a summary is logged in the Ropo One™ system to support follow-up and case management.
– Classifying a call takes about 1–2 seconds per call, while summarizing it takes from 30 seconds to one minute. Due to the scale, this adds up to a significant workload. Automating these tasks saves the equivalent of 1–2 full-time employees annually, which is a very substantial saving. It also improves job satisfaction, as staff can focus on more meaningful customer interactions. As a growing company with increasing call volumes, these benefits will only grow over time, Parkkonen says.
The first results exceeded expectations
Ropo partnered with Epical to explore how these processes could be automated. The solution was built in Microsoft Azure using OpenAI language models and other Azure services. The project demonstrated that language models can provide real, practical tools for automation.
- Even the first initial pilots performed surprisingly well — even in real conversations in Finnish. Of course, we encountered some limitations, but we found effective ways to manage them. Language models are developing rapidly and can be applied even more broadly in the future, says Petri Lindgren, Senior Consultant at Epical.
Epical played a central role throughout the project. Ropo defined the goals and requirements, while Epical translated them into a fully functioning solution.
- Epical has been easy to work with, allowing us to focus on what really matters. Their expertise has been top-notch. They’ve guided us, responded thoughtfully to our feedback, and explained clearly when something wasn’t feasible. They’ve delivered their part with excellence, and I’ve had no hesitation standing behind their work, Parkkonen says appreciatively.
Driving better reporting and smarter management
The biggest AI-driven gains at Ropo come from call summarization. Previously, during peak hours, some longer calls were not summarized adequately, which made follow-up and case tracking more difficult. Automation has solved that problem.
Ropo tested the accuracy of automation with a sample of 10,000 calls that were classified both manually and by AI. Human agents successfully classified fewer than 6,000 calls, about 60%. AI, on the other hand, correctly categorized 9,842 calls — which is nearly 100%.
– Automation saves our staff time and reduces workload, but just as importantly, it significantly improves the quality of our reporting. When we understand what the calls are actually about, we can improve our customer service and lead operations more effectively, says Parkkonen.
AI also provides reliability in generating call summaries. In service delivery, these summaries are crucial for follow-up. There’s no need to re-listen to the call later, the AIgenerated summary provides all the necessary context.
– These are excellent examples of how AI can add real value. Of course, the results depend on how well the AI is configured. Epical has done a great job: they have the expertise to fine-tune the AI so it can classify and summarize calls with high quality, Parkkonen adds.
Next stop: Message Center - encouraging others to explore AI
The success of the project has inspired Ropo to expand its use of AI – next, AI will be applied to the Message Center.
– MyRopo is our customer portal where consumers can manage their invoicing and payments. We’re adding a Message Center to the service, where AI will help respond to customer inquiries. When a customer sends a message about an invoice, AI will first analyze the message and reply accordingly. Depending on the customer’s response, the AI can continue handling the case or pass it on to a service agent. And there’s no reason why AI couldn’t eventually handle live conversations too, Parkkonen says.
Parkkonen also encourages other organizations to start exploring AI and automation.
– You don’t need a massive budget to get started, just begin by solving smaller problems within your business. For example, classifying phone calls is a small part of our overall process, but it was critical in determining whether AI could be useful to us. Once you Internal know that simple tasks can be handled well, the barrier to automating more complex processes becomes much lower, Parkkonen concludes.
