[INTERVIEW] Korean fintech uses AI to curb default rates in Indonesia

PFC Technologies (PFCT) CEO Lee Soo-hwan poses after the interview with The Korea Times at the fintech company's headquarters in Seoul, Aug. 22. Courtesy of PFCT

PFC Technologies (PFCT) CEO Lee Soo-hwan poses after the interview with The Korea Times at the fintech company's headquarters in Seoul, Aug. 22. Courtesy of PFCT

PFCT aims to become Asia's leading 'lending fintech'
By Lee Yeon-woo

Technology may not solve every social issue. But when it comes to assessing borrowers — those who have the will and ability to repay but struggle to secure loans due to their low credits — it has shown notable success.

At the forefront is PFC Technologies (PFCT). Launched in 2015 as a peer-to-peer (P2P) lending platform, the company took a major leap forward nearly a decade later, with Lee Soo-hwan stepping in as CEO in August 2023.

Under Lee's leadership, the company's competitive advantage now stems from its AI-powered credit evaluation model, AIRPACK. This B2B solution assists individuals with middle to low credit scores in avoiding excessive debt and allows financial institutions to set interest rates that mitigate delinquency risks.

Starting this year, PFCT partnered with the Indonesian subsidiaries of OK Financial Group and KB Kookmin Bank. Its aim is to lower the proportion of high-risk debtors in Indonesia by 2-3 percent.

Marking the one-year anniversary of his appointment, Lee sat down with The Korea Times to discuss the company's overseas expansions and future strategies.

Below are excerpts from the interview with Lee. They have been edited for clarity and readability.

Q. Why did you choose Indonesia for your first overseas expansion?

A. Before joining this company, I worked at a consulting firm, where we typically used a top-down approach — evaluating factors like market size and growth potential, etc. But in a start-up, you cannot operate like that. You go where the opportunity is. In our case, one of our key clients in Korea, OK Financial Group, saw our technology and asked if we could provide our solution to their Indonesian business, OK Bank Indonesia. That was our starting point.

Loan default rates in Southeast Asia are significantly higher than in Korea, and many financial institutions have struggled with traditional methods to address this issue. We began by asking whether young, creative talent could tackle this problem. After giving it a try, the results were surprisingly positive. This experience boosted our confidence and led us to realize that we could expand not just in Indonesia but across Asia as well.

PFC Technologies (PFCT) CEO Lee Soo-hwan, left, shakes hands with KB Bank Director Dodi Widjajanto during an MOU signing ceremony at KB Bank's headquarters in Indonesia, July 19. Courtesy of PFCT

PFC Technologies (PFCT) CEO Lee Soo-hwan, left, shakes hands with KB Bank Director Dodi Widjajanto during an MOU signing ceremony at KB Bank's headquarters in Indonesia, July 19. Courtesy of PFCT

Q. What was the process like?

A. We start by discussing which product to focus on first. There are a few types of loans in Indonesia — not only credit but also pension loans and motorcycle collateral loans. We spend the time discussing which areas would have the most impact on them. It's important because they need to prepare on their side as well, especially in terms of getting data ready.

After that we work on building the infrastructure. Since we're implementing AI, most financial institutions' existing systems can't handle it. So we need to help set up the infrastructure such as computer systems and internal networks. This takes the longest, sometimes up to two months.

Once that's done we work closely with the local team. To succeed, we need to understand the loan environment, lending systems, and data infrastructure of a country. It's essential to establish strong networks and business relationships with people who have a deep knowledge of these areas in that country. In Indonesia, we've built up networks with individuals who have decades of experience as bankers. It was especially helpful to receive their advice.

We move on to development afterward. This actually doesn't take long. Usually, the first version is ready within a week. So in about a week, we present the results officially to the executives. There are different versions of this, but that's the general process we followed in Indonesia.

Q. Were there any instances of failures?

A. Failures do happen. One common reason is when we fail to build strong "people-to-people" relationships. Even if the numbers look good on paper, without solid collaboration and alignment with local partners, the results can end up being practically useless. It's crucial to have a strong sense of teamwork to avoid these pitfalls.

Q. Why did you implement AI for credit scoring?

A. Over the past nine years of our journey, our default rates have consistently been low. I believe that's largely due to the work of our credit assessment team. However, as the company grew, it became impossible to solely rely on manual processes. To internalize this, we hired talented individuals to build effective models using machine learning, and we refined them further. We never saw AI as the only solution, and we still don't.

I believe that strongly sets us apart from other financial institutions using AI. Relying solely on AI experts isn't enough. They are not necessarily experts in finance or risk assessment. You need a diverse team, or else you risk ending up with solutions that we think are amazing but aren't applicable in real-world scenarios. In our case, we have more than 20 people, including credit and AI experts.

PFC Technologies (PFCT) CEO Lee Soo-hwan speaks to The Korea Times at the fintech company's headquarters in Seoul, Aug. 22. Courtesy of PFCT

PFC Technologies (PFCT) CEO Lee Soo-hwan speaks to The Korea Times at the fintech company's headquarters in Seoul, Aug. 22. Courtesy of PFCT

Q. What are the advantages/disadvantages of using AI to assess mid- to low-credit borrowers?

A. Risk assessment for mid- to low-credit borrowers is not just about traditional credit information. Alternative data is becoming increasingly important as well. And AI can definitely be used to improve this process. This way, instead of applying the highest interest rate of 20 percent to all mid- to low-credit borrowers or rejecting most of them, we can identify those with good potential and offer them better rates and conditions.

One major drawback is not knowing why a particular result was generated. When a loan is rejected, we need to provide a clear reason. Saying "AI decided" isn't a valid explanation. That's why we developed a separate model to interpret AI's results.

Q. Are there any challenges in expanding business using AI?

A. It's very, very difficult, especially due to network separation regulations in the financial industry. The best way to provide our technology is through a cloud SaaS model, where we connect to their cloud, embed our models, and manage them. But's that's impossible for most financial institutions, so we have to deliver our solutions through traditional on-premise servers.

Q. Are there other countries you are considering for expansion?

A. We're continuing to expand within Indonesia, but the next market we're most focused on is Vietnam. Many Korean financial institutions that have entered Vietnam already have strong business relationships with us. We might look at Japan next, though the exact order is still uncertain.

As for our overseas expansion strategy, we jokingly call it the "Hyundai's second-tier vendor strategy." Whenever Hyundai goes, all the second-tier parts suppliers follow. Similarly, we follow where Korean financial institutions go.

Q. What is PFCT's future ambitions?

A. Fintech as an industry is experiencing global innovation, not just in Korea. We have always questioned why our technology should be limited to Korea. What we've done over the past year is put the curiosity — whether we could monetize our technology internationally — into action.

If you ask 100 people who know our company, 99 of them would probably just say it's a P2P company. I want to break that mold. PFCT is a lending tech company. Based on that core we are carrying out various monetization strategies, including P2P lending and AIRPACK. My vision is make PFCT to become Asia's leading lending tech company.

Top 10 Stories

LETTER

Sign up for eNewsletter