Top 10 AI Prompts and Use Cases and in the Financial Services Industry in Viet Nam

AI is rapidly reshaping Vietnam's financial services: from eKYC and dynamic credit scoring to real‑time fraud monitoring and 24/7 chatbots that improve customer experience and cut operational cost, helping banks and fintechs scale inclusion and reduce non‑performing loans.

Backed by a national AI strategy, government incentives and a proposed regulatory sandbox - plus big moves such as NVIDIA's local expansion and a growing data‑center footprint - Vietnam is positioning finance to use AI at scale, even as firms must navigate new rules like the Draft Law on Digital Technology Industry and the upcoming personal data protection law.

With Google estimating AI could add VND1,890 trillion to the economy by 2030 and surveys showing roughly 80% of Vietnamese businesses adopted AI recently, the “so what” is simple: financial firms that pilot responsibly today can turn automation into measurable ROI and new products for underserved customers.

Vietnam AI regulatory roadmap and economic projections for context.

Document	Topic	Date
Decision No.127/QD-TTg	National AI Strategy (R&D & application to 2030)	Jan 26, 2021
Decree No.13/2023/ND-CP	AI/data protection alignment with global standards	Apr 17, 2023
Draft PDP Law	Personal Data Protection (affects AI, finance)	Enactment expected Oct 2025

“We shouldn't just be end-users of foreign technologies, we should create our own, by Vietnamese, for Vietnamese.” - Professor Vu Ha Van

Methodology focused on three practical lenses relevant to Vietnam: small-scale LLM prompt testing, real-world vendor case studies, and the national regulatory/implementation context.

For LLM testing, the approach mirrors the Vietnam Coracle experiment - nine prompts run across three major models - to see how chatbots handle factual queries and local nuance (Vietnam Coracle nine-prompt LLM test).

Vendor and bank examples were chosen from survey-backed reporting: Finastra's Vietnam findings and industry notes that show 94% of institutions are bullish on AI and concrete deployments such as TPBank's facial system that analyzes “128 criteria” and VietinBank's FaceID kiosks that cut processing time by about 30% (Finastra survey on AI adoption in Vietnamese financial services and bank case studies).

Finally, selection favored pilots that align with Vietnam's evolving policy and investor signals - NVIDIA's local expansion and draft laws that enable sandboxes - and with practical roadmaps and upskilling checklists so Vietnamese firms can move from pilots to measurable ROI (AI pilot-to-scale implementation roadmap for Vietnamese financial services), ensuring examples are both technically plausible and locally grounded.

Source	Role in selection	Key evidence
Vietnam Coracle	LLM prompt-testing model	Nine prompts across three LLMs
Vietnam Briefing / Finastra	Industry adoption & bank examples	94% institutions positive; TPBank 128-criteria ID; FaceID saves ~30%
Vietnam AI Sector (policy)	Regulatory & investment context	NVIDIA expansion, draft laws, regulatory sandbox

“To find out what AI says about Vietnam Coracle, I gave 9 prompts to three different LLMs (ChatGPT [Open AI], Gemini [Google] and Claude ...”

VPBank's playbook for automated customer service and conversational finance shows how Vietnamese banks can pair scale with human-centred design: migrating core workloads to the cloud unlocked big efficiency wins (28 applications moved to AWS in 11 months and the Phoenix risk job that used to take two hours now runs in about 20 minutes), while an internal automation program powered by UiPath has already answered over 1.8 million customer inquiries in 2024 and deployed 102 automation processes, cutting manual work equivalent to roughly 350 full‑time employees and running Contact Center 247 with 40+ automations handling ~150,000 transactions monthly.

That hybrid approach - cloud modernisation for speed and resilience plus RPA and conversational layers for 24/7 service - keeps advisors in the loop (automation supports, not replaces), reduces turnaround on loan and fraud workflows, and sets the stage for GenAI pilots to shorten fraud‑resolution times.

Read the full VPBank cloud story on AWS and the automation case study from UiPath for practical details and measurable outcomes.

Metric	Result
Applications migrated to AWS	28 (11 months)
Phoenix risk processing time	2 hours → ~20 minutes
Automated customer inquiries (2024)	1.8M+
Automation processes implemented (2024)	102
Contact Center automations	40+; ~150,000 transactions/month

“Adopting the AWS Skills Guild framework and launching our own enablement program called LevelUp was a logical step,” said Tran Thi Diep Anh, CHRO of VPBank.

Mastercard's real‑time transaction monitoring shows how AI can harden Vietnam's payments ecosystem without slowing everyday commerce: its Decision Intelligence and related platforms scan vast streams of transaction data to assign a risk score in about 50 milliseconds, using behavioral biometrics and networked intelligence to flag unusual patterns while cutting false declines and customer friction - a model Vietnamese banks and fintechs can tap for smoother authorizations and faster fraud response.

Global case studies report major gains (AWS says their solution helped detect three times the fraud and cut false positives tenfold), and reporting notes Mastercard now analyzes transaction volumes at scale to power adaptive, self‑learning models that learn new schemes quickly; these systems still benefit from human oversight to avoid bias and preserve trust.

Learn more about Mastercard's fraud tools and real‑time analytics in the Business Insider feature on Mastercard Decision Intelligence real-time analytics and the AWS case study on Mastercard AI/ML deployment for fraud detection.

Metric	Result / Source
Transactions scanned per year	Nearly 160 billion (Business Insider)
False declines reduced	~50% reduction reported (Bernard Marr)
Fraud detection / false positives	3× more fraud detected; false positives reduced tenfold (AWS case study)

“AI enables real-time detection of suspicious transactions by identifying patterns and anomalies impossible for human analysts to spot at scale,” - Lim, quoted in Business Insider.

Zest AI's machine‑learning underwriting offers a concrete playbook for Vietnamese lenders wanting to lift approvals without taking more risk: by ingesting hundreds of variables and alternative data, Zest turned slow, manual reviews into fast, explainable decisions - First Hawaiian Bank moved from 4% to ~55% automated decisioning (a 13× jump), delivered instant approvals for 40% of applicants and built a production model in six months, showing how a local bank can scale inclusion while freeing underwriters for complex cases (Zest AI First Hawaiian Bank machine-learning underwriting case study).

Independent reporting frames the payoff bluntly: AI can turn a grainy credit score into “high‑definition, 3D” borrower insight and has been associated with roughly a 25% lift in approvals at adopters while holding portfolio risk steady (Quartz analysis of Zest AI loan approval gains and borrower insights).

For Viet Nam, the “so what” is practical - automated, fair, and auditable models can expand credit to thin‑file customers, speed decisions at digital touchpoints, and slot into pilot‑to‑scale roadmaps that regulators and banks are already discussing.

Metric	Result (Source)
Automated decisioning	4% → ~55% (13× increase) - Zest AI case study
Instant approvals	40% (9× increase) - Zest AI case study
Time to production	6 months to launch - Zest AI case study
Reported lift in approvals	~25% increase (adopters) - Quartz

“Zest AI's underwriting technology is a game changer for financial institutions. The ability to serve more members, make consistent decisions, and manage risk has been incredibly beneficial to our credit union.” - Jaynel Christensen, Chief Growth Officer (Zest AI)

For Vietnamese banks and fintechs facing tighter AML/KYC scrutiny, Denser compliance stacks - built from RAG-grounded LLMs, document automation and hybrid GraphRAG architectures - make regulator response and ongoing monitoring practical at scale: systems can ingest mixed-format circulars, chunk a 200‑page regulation into concise, auditable summaries, and surface the exact clauses that map to internal KYC checks so teams can answer regulator queries in hours instead of weeks (and show the source).

Practical patterns from recent work include retrieval-augmented pipelines to ground answers in law texts, knowledge graphs to trace obligations across policies and controls, and domain fine‑tuning so models understand local banking jargon; together these reduce manual review while preserving human-in-the-loop checkpoints and full audit trails.

Vendors and enterprise pilots also emphasise alignment and guardrails - refusing uncertain answers, logging sources, and routing edge cases to compliance officers - so automation improves speed without sacrificing accountability.

For a practical playbook Viet Nam teams can follow, see guidance on parsing and summarizing regulatory documents and a Vietnam-focused pilot-to-scale roadmap to prove ROI and governance.

“It makes sense to leverage AI where our documents and knowledge get stored: our DMS.” - Chris Jaglowitz, Common Ground Condo Law (quoted in NetDocuments)

FPT Software brings a practical playbook for Vietnamese banks and fintechs wanting to turn paperwork and legacy systems into real-time operational capability: intelligent document processing (IDP) plus RPA turns mixed PDFs, images and multi‑language forms into structured data that feeds core banking, reconciliation and KYC workflows, and a recent FMCG case processed 21,000 documents (4,000+ pages in five languages) and delivered measurable improvements in about three months - a useful fast‑win for finance teams overloaded with loan docs and invoices.

Coupled with FPT AI Agents, which automate invoice processing, financial reporting and multi‑channel customer care while “Vietnamizing” interfaces and knowledge, the stack can cut repetitive work dramatically (IDP+RPA patterns report up to an 80% reduction in manual data entry) and free analysts to handle exceptions and governance.

For institutions planning pilots, the evidence is clear: start with IDP for back‑office hot spots, plug agents into ERP/CRM, and use proven tooling for code and security hardening as FPT has done in its development lifecycle.

Explore FPT's IDP case study and FPT AI Agents for implementation details and local examples.

Metric / Capability	Result / Source
Documents processed (case study)	21,000 documents; 4,000+ pages in 5 languages (FPT IS IDP case study)
Time to implement improvements	3 months (FPT IS IDP case study)
Manual data entry reduction (IDP + RPA)	Up to 80% reduction (FPT IS)
Security & quality tooling	~200 projects/year using Coverity SAST & Black Duck SCA (FPT / Black Duck case study)

“We took the path of looking into tools to improve code quality and security as early as possible in the development lifecycle.” - Do Van Khac, Chief Delivery Officer, FPT Software

BloombergGPT's finance‑first approach - a 50‑billion‑parameter LLM trained on Bloomberg's decades of market text - offers Vietnamese banks and research teams a ready example of how generative AI can accelerate analysis, reporting and desk‑level research: it can summarize news, generate headlines and earnings writeups, perform sentiment analysis and named‑entity recognition, and even translate plain‑language queries into Bloomberg Query Language for structured retrieval.

For Vietnam firms that need accuracy and auditability, the practical play is clear: pair a domain model like BloombergGPT with retrieval‑augmented pipelines or fine‑tuning on local datasets so outputs are grounded in Vietnamese filings, local news and internal data - an approach explained in the LLM fine‑tuning guide - and consider smaller, task‑scoped models or PEFT to cut costs while preserving performance.

For a deeper read on the model and best practices for adapting LLMs to finance, see the BloombergGPT finance‑focused LLM overview and the LLM fine‑tuning guide for enterprises.

Attribute	Detail
Parameters	50 billion
Financial training tokens	~363 billion (Bloomberg)
General training tokens	~345 billion
Core capabilities	News summarization, sentiment analysis, NER, BQL conversion

“Writing a really great prompt for a chatbot persona is an amazingly high‑leverage skill and an early example of programming in a little bit of natural language.” - Sam Altman

For Vietnamese asset managers, insurers and pension funds looking to move beyond siloed spreadsheets, BlackRock's Aladdin acts as

a common data language

that can unify public and private holdings, risk analytics and trading workflows into a single operating model - making it easier to spot exposures, test scenarios and scale systematic strategies without rebuilding plumbing from scratch; pairing Aladdin's platform with BlackRock's AIM approach lets teams combine hundreds or thousands of signals into customizable, regime‑aware alpha models while keeping explanations and portfolio simulators front and centre so decisions remain auditable and risk‑aware (see the Aladdin platform and BlackRock's writeup on Augmented Investment Management).

For Vietnam that means a practical route to build custom indices, lift private‑market visibility after BlackRock's Preqin tie‑ups, and pilot ML‑driven portfolio construction with a clear pilot‑to‑scale playbook - because, as BlackRock notes, even a small, repeatable forecasting edge can compound materially over time.

Ready-to-run pilots and local governance checklists help prove ROI before scale-up; follow a Vietnam-focused pilot roadmap to start responsibly.

Capability	Why it matters for Viet Nam
BlackRock Aladdin whole-portfolio integration platform	Unifies public & private holdings, risk and operations for clearer exposures
BlackRock AIM Augmented Investment Management for custom alpha models	Machine‑learning pipelines to combine many signals into explainable forecasts
AI pilot-to-scale roadmap for Vietnam financial services	Practical playbook to prove ROI and governance before broad adoption

Techcombank is turning its “data brain” and cloud-first platform into a personalization engine that makes marketing feel like timely advice rather than noise: by analysing roughly five billion daily data points and piloting a personalised digital agent and call‑centre tone‑analysis bots, the bank can surface AI‑driven spending insights, nudges and product offers at the exact moment a customer needs them - a strategy that helped drive a reported 20.3% rise in profit before tax in 2024 and attracts millions of digitally acquired customers.

Early experiments show impact: a three‑week Personetics pilot for 10,000 users lifted savings balances by 9%, installment volume by 43.7% and monthly log‑ins by 444%, while a Backbase‑powered mobile platform (with AI PFM, conversational flows and NFC onboarding) supports rapid scale - 1.9 million digitally acquired customers and a top app rating - turning transactional data into measurable growth and deeper customer loyalty in Vietnam's competitive market (Techcombank digital banking case study - Backbase, Personetics personalized banking pilot results, Techcombank growth and innovation profile - Asian Banker).

Metric	Figure / Source
Profit before tax (2024)	+20.3% (Asian Banker)
Daily data points analysed	~5 billion (Asian Banking & Finance)
New digital customers (impact)	1.9M (Backbase)
Personetics pilot (10,000 users)	Savings +9%; Installments +43.7%; Log‑ins +444% (Personetics)
Mobile app rating	~4.9/5 (Backbase)

“AI would be the ‘cornerstone' of Techcombank's transformation.” - Alexander Macaire, CFO (quoted in Asian Banking & Finance)

VinAI's research pedigree and recent deal-making put Vietnam on the map for AI-driven underwriting and pricing: the Ho Chi Minh–based firm - known for generative AI, machine learning, computer vision and NLP and part of the Vingroup ecosystem - sold its generative AI research business to Qualcomm in April 2025, a signal that homegrown capability is attracting global partners (A&O Shearman coverage of the VinAI–Qualcomm transaction).

That matters because the Vietnam market for AI in model insurance is set to explode - from USD 4.1 billion in 2025 to USD 14.7 billion by 2031 - creating real demand for predictive engines that can price risk faster, tailor premiums to behaviour and cut fraud investigator hours (Vietnam AI in Model Insurance Market forecast and growth).

Recent commercial launches, like SCOR's predictive engine for Vietnamese life business, show how reinsurer-grade models can be adapted locally and point to practical partnerships where VinAI's tech and talent could accelerate automated underwriting pilot-to-scale programs (SCOR L&H AI-based predictive engine launch in Vietnam), turning underwriting into a faster, more personalised customer touchpoint rather than a paperwork bottleneck.

Fact	Detail / Date
Vietnam AI in Model Insurance market	USD 4.1B (2025) → USD 14.7B (2031); CAGR 23.6%
VinAI corporate milestone	Sold generative AI research business to Qualcomm - Apr 7, 2025
VinAI capabilities	Generative AI, ML, computer vision, NLP (per A&O Shearman)

“This transaction is a clear indicator of the interest and investment in AI technologies within the ASEAN region.” - Tina LeDinh, A&O Shearman

Viettel's VCS Threat Intelligence brings a Vietnam‑ready layer to identity, access and insider‑risk defense by turning noisy signals into actionable early warning: AI and machine‑learning filter raw feeds from the clear, deep and dark web, then experts verify hits and push real‑time alerts about new vulnerabilities, ransomware, APT activity, leaked credentials or fake websites that could impersonate a bank; the service also surfaces unknown or unmanaged systems and unusual behaviour on public‑facing assets so teams can stop escalation before customer data is exposed.

For financial firms building detection pipelines, VCS is designed to plug into SIEM/IPS workflows and deliver findings via portal, notifications or API, while offering takedown support and tailored reports that prioritise what regulators and risk teams in Viet Nam need to see.

See Viettel's Threat Intelligence overview for feature details and a practical primer on anomaly‑detection methods for network and insider threats from the Network Anomaly Detection guide.

Capability	What it delivers
Real‑time alerts	Vulnerabilities, malware/ransomware, APTs, leaked credentials, fake sites, unusual asset behaviour
AI/ML processing	Automated filtering, analysis and organisation of threat data from multiple sources
Expert verification & support	Analyst review, investigations, takedown assistance and tailored remediation advice
Delivery & integration	Portal, notifications, API - integrates with SIEM/IPS/ticketing systems for fast response

“You can't defend against what you don't see.”

Start small, start governed: Vietnamese banks and fintechs should run focused pilots in the new regulatory sandbox (Decree 94/2025) to prove value, then use a clear pilot‑to‑scale implementation roadmap to lock in ROI and controls; pair each pilot with board‑level AI governance and a “minimum viable” oversight package - explainability, human‑in‑the‑loop, audit trails and incident playbooks - so innovation doesn't outpace trust.

Invest in people now (Vietnam needs roughly 400,000 more tech workers by 2026) and map hires or training to specific use cases - customer chatbots, IDP for loan docs, or AML monitoring - so skills directly enable pilots; use upskilling checklists and bootcamps to accelerate readiness and close gaps.

Coordinate with regulators and peers, surface ethics through an internal AI Ethics Council, and document every pilot so lessons scale rather than being re‑built.

For practical next steps, explore the Vietnam AI policy and sandbox overview in the Vietnam News coverage and follow a tested pilot‑to‑scale implementation roadmap, while using the practical upskilling checklist to turn jobs‑at‑risk into new, higher‑value roles - because in Vietnam the race is not to automate fastest, but to govern and upskill smartest.

“The application of AI is not only a trend, but also an urgent requirement for the Vietnamese banking system to improve its competitiveness, provide modern services and ensure the safety and stability of national finance and currency.” - Dr Nguyễn Quốc Hùng