How Kash automates 90% of support across complex fintech journeys with Fin

Kash is a mobile app that puts a full bank in your pocket across Central America. Part of BAC International Bank, it lets people open and manage accounts, move money, pay bills, and take out microloans from their phones, with transactions running over Visa, Mastercard, and local payment networks. When money and trust are on the line, a slow or uncertain answer is not a small thing.

For years, support ran on people alone. Every question, from a forgotten password to a failed transfer, landed with a human agent. That approach worked while Kash was smaller, but as its user base expanded across the region and support volume climbed to more than 20,000 conversations a month, the model began to reach its limit. The challenge was not just the volume, but the work each conversation required. Most involved multi-step financial issues that required customer context, troubleshooting, and careful handling. Many followed recurring patterns, but agents still had to work through every case manually from start to finish.

Adding more agents was not sustainable, and the team could not flex quickly enough when volumes spiked. On the busiest days, customers waited longer, while this constant flow of manual work took time away from the sensitive cases that genuinely required human judgment.

Kash needed automation it could trust with complex, regulated financial workflows, while keeping full control over how every customer interaction was handled. So they built toward it in stages rather than flipping a switch. Starting in 2023, they leaned on automated workflows to take on about half of their conversations, learning what could be handled without a person well before AI entered the picture.

In September 2025, they launched Fin, an AI Customer Agent that now sits at the front of every conversation, backed by a detailed knowledge base and live connections into their systems. The real shift was depth. Using Procedures, the team taught Fin to resolve complex, multi-step queries end to end. When a transfer fails, Fin checks live transaction data, applies the relevant policy, identifies what went wrong, and tells the customer exactly what will happen next and when. Rather than handing them off, it resolves the issue in the moment. Fin carries out that same end-to-end process across channels, whether customers reach out inside the Kash app, on Facebook, or on Instagram, while adapting its tone to each market: more formal in Costa Rica and more relaxed elsewhere in the region.

Just as importantly, Kash stays in control of the whole experience. They've set clear handoff rules that tell Fin when to route a conversation to a human the moment it needs judgment or sign-off, keeping the highest-risk decisions with the team.

Today, Fin is the first point of contact for every Kash customer conversation, automating more than 90% of total volume, so fewer than 10% ever reach an agent. And that scale hasn’t come at the expense of quality: CSAT has remained close to 90% throughout the rollout, with customers now receiving instant, personalized support 24/7.

The change reshaped the team. Agents who once cleared tickets now spend their time making Fin better, refining its guidance, sharpening the knowledge base, and extending Fin into more complex, multi-step cases, from failed transfers and security checks to phone number changes and account recovery. Support is no longer about keeping pace with demand. It's now a continuously improving system that the team can shape, strengthen, and scale with the business.

"Our objective went beyond conversation automation; we wanted to create an experience where customers could resolve their needs quickly and effectively," says José Manuel Páez, CEO at Kash. "The combination of Fin, a robust knowledge base, and agile integrations transformed support into a highly scalable operation while maintaining an excellent customer experience."