How Lenders in Vietnam Reduce Loan Defaults with AI Calling
Lenders in Vietnam reduce loan defaults with artificial intelligence (AI) calling by reaching every borrower early, in natural Vietnamese, before an account slides into arrears. Voice AI agents deliver payment reminders, confirm intent, offer restructuring options, and escalate hardship cases to humans — cutting delinquency while staying inside State Bank of Vietnam conduct rules and using credit-bureau data responsibly.
Why Do Lenders in Vietnam Struggle With Defaults?
Consumer finance is concentrated and competitive. Three players — FE Credit, Home Credit, and Mcredit — hold over 80% of the consumer-finance market, lending to millions of thin-file and first-time borrowers (Global Referral, 2024). Many of these customers are hard to reach and easy to lose to delinquency if contact is late or inconsistent.
Credit-risk infrastructure is maturing to help. The National Credit Information Center (CIC), a public body under the State Bank of Vietnam (SBV), holds credit data on more than 30 million borrowers — roughly a third of the population — with reporting from effectively all credit institutions (FPT IS). Alongside it, the private Vietnam Credit Information Joint Stock Company (PCB), with CRIF, operates the country's first private credit bureau under SBV licence (CRIF, press release).
Better data identifies who to call; AI calling makes reaching them at scale affordable — the same logic behind AI voice agents for loan collections.
How Does AI Calling Reduce Defaults Step by Step?
Default prevention is a timing game. AI lets lenders act in the pre-delinquency and early-bucket windows, where cure rates are highest.
- Pre-due reminders. The agent calls before the due date, confirms the amount in ₫, and shares payment options — preventing accidental misses.
- Day-past-due follow-up. For missed payments, the agent calls promptly, understands the reason, and captures a promise-to-pay.
- Restructuring signposting. For genuine hardship, the agent explains available options and warm-transfers to a human officer.
- Segmented cadence. Higher-risk accounts, flagged using CIC or PCB data, receive earlier and more frequent, but always respectful, contact.
- Escalation with context. Complex or vulnerable cases route to human collectors with a full transcript.
Early contact matters most — see voice AI for early-bucket debt resolution.
Delinquency stage | Goal | AI calling role |
|---|---|---|
Pre-due | Prevent misses | Proactive reminders |
1–30 days past due | Cure early | Prompt follow-up, promise-to-pay |
31–60 days past due | Restructure | Signpost options, escalate |
60+ days past due | Human-led recovery | Context handoff to officers |
What SBV Conduct Rules Must AI Calling Follow?
Lending and collections conduct sits under SBV regulation. Circular 39/2016/TT-NHNN governs lending transactions of credit institutions and foreign bank branches, and was tightened by Circular 06/2023/TT-NHNN, effective 1 September 2023 (Lexology, 2023). Collections practices must be professional — no threatening, harassing, or deceptive conduct.
AI calling supports compliance in three ways: it keeps a consistent, pre-approved script; it records every call for audit; and it enforces contact-frequency and time-of-day limits automatically. Borrower data drawn from CIC or PCB must be processed under Decree 13/2023/ND-CP and the Personal Data Protection Law (91/2025/QH15, effective 1 January 2026).
This is an explainer, not legal advice. Lenders should validate scripts, cadence, and consent with compliance before deployment. For a broader framework, see the AI banking collections playbook.
How AI Helps
YuVoice runs natural Vietnamese voice agents that reach every borrower in the pre-delinquency and early-bucket windows, where cures are cheapest. It delivers reminders, captures promises-to-pay, signposts restructuring, and escalates hardship cases to human officers with full context. Because each call follows a fixed, compliant script and is recorded, lenders get a defensible audit trail against SBV conduct expectations. Risk data from CIC or PCB can prioritise who to call first, so scarce human collectors focus on complex, higher-balance cases. The result is lower roll-forward into hard buckets — without harassing borrowers or expanding headcount.
FAQ
Q1. Does AI calling actually lower default rates? By contacting borrowers early and consistently — before and just after the due date — AI calling lifts cure rates in the buckets where recovery is cheapest, reducing roll-forward into serious delinquency.
Q2. Is AI collections calling allowed in Vietnam? Collections must follow SBV conduct rules and lending regulations such as Circular 39/2016 as amended by Circular 06/2023. AI can call compliantly by using approved scripts, recording calls, and enforcing contact limits. This is not legal advice.
Q3. How does CIC data fit in? The National Credit Information Center provides credit data used to assess and prioritise risk. Lenders can use it to decide which borrowers to contact earlier, while processing that data under Decree 13/2023 and the 2025 Personal Data Protection Law.
Q4. What about borrowers who genuinely cannot pay? The agent identifies hardship, avoids pressure, explains restructuring options, and warm-transfers the case to a human officer with a full transcript for a fair resolution.
Q5. Can AI calling harass borrowers? It should not. Well-configured systems enforce respectful scripts, time-of-day windows, and frequency caps — reducing the harassment risk associated with poorly supervised human or third-party collectors.
Q6. Which lenders benefit most? Consumer-finance companies and banks with large, thin-file portfolios benefit most, because AI makes early, high-volume, low-cost contact economically viable across millions of small-ticket loans.
Conclusion
Defaults are cheapest to prevent early — and that is exactly where human-only calling cannot scale. AI calling lets Vietnamese lenders reach every borrower on time, in Vietnamese, with compliant, recorded conversations, while reserving human collectors for hardship and high-balance cases. Paired with CIC and PCB data, it turns timing into lower losses.
Reduce loan defaults with compliant AI calling. Talk to the YuVerse team
References
- Global Referral — An Overview of Vietnam's Consumer Finance Market — https://globalreferral.group/an-overview-of-vietnams-consumer-finance-market/
- FPT IS — Data Management System, Vietnam National Credit Information Center (CIC) — https://fpt-is.com/en/customers/data-management-system-vietnam-national-credit-information-center-cic/
- CRIF — PCB and CRIF obtain a licence to develop the first private credit bureau in Vietnam — https://www.crif.com/knowledge-events/press/pcb-and-crif-officially-obtain-a-license-to-develop-the-first-private-credit-bureau-in-vietnam/
- Lexology — State Bank of Vietnam amends Circular 39 on lending transactions — https://www.lexology.com/library/detail.aspx?g=0cb80123-d163-4242-a0d4-efea85014e13
