Kaspay: Revolutionizing Customer Support with AI-Powered Solutions

Customer Challenge

Kaspay transaction volumes and customer base expanded rapidly, their support infrastructure faced mounting pressure.

"Our success created a unique challenge," explains Maria Sanchez, VP of Customer Experience at Kaspay. "As we scaled to serve more enterprise clients across our diverse product portfolio, our support team was struggling to maintain the high service standards that our customers expect."

The support team faced several interconnected challenges:

Growing Support Volume vs. Team Capacity With each new enterprise client and product addition, support requests increased exponentially. The traditional approach would have required significant headcount expansion a solution that wasn't aligned with Kaspay's efficiency-focused culture.

"We didn't want to solve a scaling problem by simply throwing more people at it," notes Sanchez. "We needed a smarter approach that would allow us to handle growing volumes while maintaining our lean organizational structure."

Inefficient Ticket Routing One of the most significant bottlenecks was the routing of support tickets. With Kaspay's extensive range of products, features, and services, ensuring that each inquiry reached the most qualified technical expert was challenging.

Knowledge Access and Response Quality Even when tickets reached the right team, support agents faced another challenge: quickly accessing the relevant information needed to provide accurate responses across Kaspay's extensive product ecosystem.

"Our support agents are incredibly knowledgeable, but no single person can be an expert on every aspect of our constantly evolving platform," says Chen. "They were spending too much time searching for information across various knowledge bases instead of focusing on solving customer problems."

As a result, resolution times were increasing, and the support team was showing signs of burnout as they struggled to keep pace with the growing demand while maintaining quality.

Upsonic AI Solution

To address these challenges, Kaspay established a specialized team of data scientists and machine learning engineers at their new Tech Hub in Madrid. This team was tasked with leveraging AI to transform their support operations without compromising quality or requiring significant headcount increases.

After evaluating various approaches, the team decided to build two complementary AI applications using Upsonic's framework:

Smart Ticket Routing System The first application focused on automatically analyzing incoming support tickets and routing them to the most appropriate technical experts based on content analysis.

The smart routing system:

• Analyzes the theme and technical complexity of each ticket

• Evaluates sentiment to identify urgent or high-priority issues

• Dynamically adjusts priority based on the customer profile

• Routes tickets directly to the most qualified technical specialist based on expertise mapping

Support Agent Copilot The second application focused on empowering support agents with instant access to the most relevant information needed to resolve customer inquiries.

Working with Upsonic, the team developed a sophisticated question-answering system that:

• Aggregated documentation from multiple sources (public and private)

• Stored documents in a vector database with an optimized embedding model

• Retrieved the most relevant and current information for each inquiry

• Generated suggested responses that agents could quickly review and customize

Why Upsonic? Kaspay chose Upsonic as their development framework for several strategic reasons:

"Upsonic provided us with a single, customizable framework that allowed us to move quickly from prototype to production," explains Moreno. "Most importantly, it gave us the flexibility to experiment with different models without getting locked into a single provider, which was crucial for our continuous improvement approach."

The technical implementation involved:

• Creating a custom class extending from Upsonic's base LLM class to ensure seamless interactions with their internal LLM API endpoint

• Integrating the system with Kaspay's event-driven microservice architecture hosted in a Kubernetes cluster

• Using Upsonic to evaluate performance and compare how different underlying models affected response quality and costs

Results and Impact

Within four months of implementation, Kaspay's AI-powered support system delivered impressive results across multiple dimensions:

Enhanced Efficiency and Accuracy

• 83% reduction in ticket transfers between teams

• 91% of tickets now routed correctly on the first attempt

• Average response time decreased by 62%

"Tickets now go directly to the right expert the first time, eliminating the delays and frustration caused by multiple transfers."

Improved Agent Experience and Productivity

• 40% reduction in time spent searching for information

• 78% of agent-crafted responses now utilize the AI-suggested content

Scalable Support Operations

• 215% increase in ticket volume handled without additional headcount

• Cost per ticket resolution reduced by 47%

• Agent focus shifted to complex problem-solving and relationship building

"Instead of a linear relationship between growth and headcount, we now have a scalable model where technology handles the increasing volume while our team focuses on the high-value work that requires human judgment and creativity."

Future Vision

Building on this success, Kaspay is now exploring additional applications of Upsonic's framework across their business:

Proactive Support Intelligence The next phase involves developing predictive capabilities to identify potential issues before they impact customers, allowing the support team to shift from reactive to proactive engagement.

Expanded Self-Service Options Kaspay plans to extend its AI capabilities to customer-facing knowledge bases and support portals, enabling more self-service options while maintaining quality and accuracy.

Cross-Functional AI Applications The success of the support team's implementation has inspired other departments to explore similar applications, from product development to compliance monitoring.