Innovator of the Year: Kakao Bank Bets on AI-Native Banking
Kakao Bank is rebuilding its service model around what it calls AI-native banking: simpler, more intuitive, and conversational financial services. At Money20/20 Asia in Bangkok, CEO Yun Ho-young put it bluntly: "AI-optimized UI and UX, backed by strong data capabilities, will determine who survives. The industry and its ecosystem will reorganize around companies with AI competitiveness." Money20/20 Asia
From Experiments to Shipped Features
In May, the bank launched AI Search, Korea's first conversational search for banking. Customers can ask natural-language questions like "How do I increase my transfer limit?" or "What are today's exchange rates?" and get fast, specific answers without digging through menus.
In June, Kakao Bank rolled out an AI Financial Calculator that autofills inputs and computes savings plans, loan payments, and currency conversions. By September, AI Search and the Calculator hit 1 million users; today they reach about 1.7 million users.
Generative AI in the Transaction Flow
The latest release, AI Transfer, applies generative AI to a core banking function. Customers describe the transfer in plain language; the assistant handles the steps, verifies details, and completes the transaction with control checks in place.
Next up is AI Group Manager, slated for December. It automates group-account tasks like tracking club dues and splitting shared expenses, with room to expand into collaborative money tools.
Building the Operating System for AI Banking
Kakao Bank created an AI Group to unify teams and speed delivery across customer services, internal processes, and global rollouts. The company operates a financial technology research center (since 2021) and an AI-specialized data center launched last year to advance proprietary tech and security.
One practical output: an AI smishing text verification service that attracted over 210,000 users in three months. Strong adoption here signals that customers will opt in when AI clearly reduces risk and effort.
Expansion: Indonesia and Thailand
Superbank in Indonesia-backed by Kakao Bank and Grab-turned profitable roughly a year after launch and has around 3 million users. In June, Kakao Bank and Thailand's SCBX won approval to establish a virtual bank, targeting a rollout in the second half of 2026.
The strategy is consistent: export the AI-native model, lead with conversational banking, and use data advantage to compress acquisition costs and improve unit economics.
What Finance Leaders Should Watch
- Conversational interfaces move from support to primary UI. Track completion rates, containment, handoff quality, and time-to-resolve as core KPIs.
- Risk and controls: combine generative AI with deterministic rails for transactions, enforce step-up authentication, keep full audit trails, and align with KYC/AML policies.
- Data stack: feature stores, clear PII boundaries, retrieval layers for policy, and continuous model monitoring for drift and bias.
- Cost levers: automate high-volume service requests and back-office workflows to improve cost-to-income ratio without sacrificing CSAT.
- Security: user-facing protections (e.g., smishing checks) build trust and reduce fraud loss.
Practical Moves for Banks and Fintechs
- Start with the top 20 intents by volume (limits, transfers, rates, bills) and design for end-to-end completion, not chat for chat's sake.
- Instrument everything: prompts, outcomes, errors, handoffs, and customer feedback. Use this to prune features that don't move KPIs.
- Blend AI with rules: strict guardrails for money movement; flexible generation for search, help, and calculators.
- Build privacy by design: role-based access, data minimization, synthetic data for training, and red-teaming for prompt-based attacks.
- Upskill teams across product, data, and compliance so models, UX, and controls evolve together.
Kakao Bank isn't treating AI as a side project. It's rebuilding core workflows, measuring adoption at scale, and pushing the model into new markets. For finance teams, the bar has moved: conversational, compliant, and measurable-or customers will pick the provider that is.
Looking to upskill your team on applied AI for finance workflows and tools? Explore curated resources here: AI tools for finance.
Your membership also unlocks:
