By Vinnie Lauria, Co-founder and Managing Partner, Golden Gate Ventures
Banks around the world are moving quickly to adopt artificial intelligence (AI). They are testing copilots, automating compliance processes, improving fraud detection, redesigning customer service and exploring how AI can support everything from credit decisions to wealth management.
Yet as access to AI becomes easier, a more difficult challenge is emerging. Two banks can deploy similar models and produce very different results. One integrates AI into daily workflows and begins improving productivity, decision-making and customer service. The other accumulates pilots that generate attention but never meaningfully change how the institution operates.
The difference is rarely access to technology. Increasingly, it is adoption. Competitive advantage will not come simply from having the most sophisticated AI model. It will come from an institution’s ability to persuade people across departments, markets and cultures to trust AI, use it appropriately and change established ways of working.
That makes AI adoption not only a technological challenge, but also a cross-cultural one.
That makes AI adoption not only a technological challenge, but also a cross-cultural one.
The same technology, different outcomes
Over the past two decades, I have invested across Asia and, more recently, the Middle East. Golden Gate Ventures has backed financial-technology (fintech) companies, including Stripe, Xendit and Funding Societies. We have also worked closely with Asian banks and insurance companies that have invested in our funds and engaged with our portfolio.
Across these relationships, one pattern has remained consistent: Technology can move across borders far more quickly than behavior. A digital product can be launched in a new country almost instantly. That does not mean customers will trust it, employees will promote it, or an organization will reorganize itself around it.
We saw this during earlier waves of fintech adoption. A product that gained traction quickly in one market could struggle in another, even when the technical capabilities were identical. Local payment habits, regulatory structures, consumer expectations and attitudes toward trust all shaped the outcome.
AI is following the same pattern, but the challenge is even more complex. It is not merely another product being introduced to customers. It is a new participant in the organization itself. AI may recommend decisions, communicate with clients, interpret internal information, assess risks and influence how employees perform their jobs. That places it directly inside the cultural fabric of a bank.
Banking has always been cultural
Banking may appear highly standardized from the outside. Financial statements, risk calculations, compliance processes and payment systems follow common principles. But banking is also built on deeply local ideas about trust, authority, privacy, relationships and risk.
In some markets, customers rapidly adopt new financial tools when the product is convenient and backed by a credible institution. In others, trust develops more gradually and depends heavily on human relationships.
Within institutions, decision-making also varies. Some organizations encourage employees at every level to experiment, question established processes and introduce new tools. Others are highly hierarchical, with employees reluctant to act until senior leadership has clearly approved a change.
Communication styles differ as well. In a direct culture, an employee may openly say that an AI tool is unreliable or that a pilot is failing. In a more indirect culture, concerns may be expressed softly, privately or not at all. A leadership team may interpret polite agreement as enthusiasm, only to discover that the technology is barely being used.
These distinctions matter because adoption depends on behavior that is often invisible in a technology plan.
A bank can purchase licenses, run training sessions and establish an AI steering committee. But none of those steps guarantee that a relationship manager will trust an AI-generated recommendation before speaking with an important client or that a compliance officer will feel safe relying on an unfamiliar system.
AI must become multicultural, not just multilingual
The cultural challenge is not limited to how people adopt AI. It also affects how AI itself behaves.
One of my more interesting observations over the past year has come through our investment in Arabic.ai. We have learned that many international companies entering the Gulf and wider MENA (Middle East and North Africa) region need something far deeper than language translation. They quickly realize their existing AI models can speak Arabic, but they do not necessarily understand the business norms, communication styles, decision-making processes, regulatory expectations or cultural context of the region. They are not looking for translation. They are looking for localization.
In our book, Mind the Gap: Scaling Businesses Across Cultures (co-authored also by Stefano Pellegrino), Professor Savanid (Nui) Vatanasakdakul of Carnegie Mellon University in Qatar introduced the concept of “transcreation” to describe this distinction. Translation changes the words. Transcreation adapts the product or experience so that it carries the right meaning and fits how people in a particular culture communicate, behave and make decisions. That concept is increasingly relevant to AI.
An AI assistant developed around Western communication patterns might be clear, direct and technically accurate. Yet the same response could appear overly blunt or inappropriate in a culture in which communication is more contextual, formal or sensitive to hierarchy.
A model may understand the literal meaning of a question while missing what the user is actually asking. It may recommend a direct escalation when the more effective local approach would be to preserve the relationship and address the issue indirectly. It may fail to recognize when politeness signals uncertainty rather than agreement.
This is why the next stage of AI development cannot be only multilingual. It must also be multicultural.
For banks, this will be particularly important. Financial institutions operate through trust, and trust is not granted in the same way everywhere. Models used in customer service, financial advice, collections, credit assessment and relationship management will need to reflect local expectations without compromising the institution’s standards, governance or ethics.
Cross-border banking raises the stakes
The need for culturally intelligent AI is growing as banking becomes more cross-border. Consumers increasingly expect to travel, work, invest and transact across countries without encountering the traditional friction of national banking systems. Small businesses can begin serving international customers almost immediately. Employees receive income across jurisdictions. Families send money across borders. Companies expect treasury and payment infrastructure to follow their operations into new markets.
Fintech companies have helped reset these expectations. Consumers now compare their banks not only with other banks, but also with the best digital experiences they receive from any financial platform. Yet banking rails, regulations, data systems and organizational structures remain largely national. Even global banks often operate as collections of country businesses connected through a common brand and governance structure.
AI could help bridge some of these gaps. It can make complex products easier to understand, improve cross-border customer service, identify patterns across fragmented systems and help employees navigate different regulatory environments.
But an AI system cannot effectively bridge markets if it assumes that customers and employees behave identically in each one. A customer in Indonesia, Singapore, Saudi Arabia or the United States may have very different expectations about privacy, financial advice, human interaction and institutional authority. Even within a single country, generational, professional and regional subcultures can shape adoption.
Global deployment, therefore, requires more than a standardized model. Banks need global clarity on governance, risk and institutional values, combined with local flexibility in how AI is implemented and communicated.
Adoption begins inside the institution
Before banks can use AI to serve culturally diverse customers, they must address the cultural barriers inside their own organizations.
Employees may worry that AI will replace them. Managers may privately resist tools that weaken their control over information or decision-making. Technology teams may measure technical performance, while business teams remain unclear about how the tool improves their work. These are not model problems. They are leadership problems.
The institutions that succeed will make AI adoption a shared organizational responsibility rather than an isolated technology initiative. Four actions are particularly important.
The institutions that succeed will make AI adoption a shared organizational responsibility rather than an isolated technology initiative. Four actions are particularly important:
Measure adoption, not pilots.
Banks often highlight the number of AI experiments underway. But the number of pilots says little about whether behavior has changed. Leadership should measure active use, repeated use, workflow integration, time saved, decisions improved and customer outcomes. They should also examine where adoption is weak and ask why.
A pilot demonstrates that a technology can function. Adoption demonstrates that an institution can change.
Build internal AI champions.
Employees are often more persuaded by trusted colleagues than by corporate mandates. Banks should identify credible people within business units who can demonstrate practical uses of AI, share mistakes and help colleagues build confidence. These champions should not sit only in technology teams. They should include relationship managers, risk professionals, operations leaders, compliance officers and country executives.
In hierarchical organizations, visible support from senior leadership will also be essential. Employees must know that thoughtful experimentation is authorized, not merely tolerated.
Align incentives.
Organizations frequently announce innovation as a priority while continuing to reward established behavior. If employees are measured only on short-term output, they may view learning a new system as a distraction. If managers benefit from controlling information, they may resist tools that distribute knowledge more widely. If a failed experiment damages someone’s career, experimentation will stop.
Leaders must align performance measures, recognition, training and promotion with the behaviors they want AI adoption to produce.
Give country teams flexibility.
A global bank needs clear standards for data security, model risk, compliance and ethics. These should not vary casually from market to market. Implementation, however, should not be dictated entirely from headquarters.
Country teams understand how local employees communicate, how customers build trust and how decisions actually move through the organization. They should have room to adapt training, interfaces, workflows and use cases within the bank’s global guardrails.
This is not a choice between central control and local autonomy. The objective is global consistency in principles and local relevance in execution.
The real competitive advantage
AI models will continue to improve. They will become more capable, more affordable and more widely available. As that happens, access to AI will become less distinctive. The real advantage will belong to banks that can turn widely available technology into trusted, repeated behavior across complex organizations.
Those institutions will understand that trust moves at different speeds, that hierarchy changes how new tools are accepted and that communication styles affect whether employees raise concerns or quietly disengage. They will recognize that localization must go beyond language and extend into the behavior of the AI itself. Most importantly, they will treat adoption as a leadership discipline rather than the final stage of a technology deployment.
AI may be global. But adoption will always be local.
Vinnie Lauria is a venture capitalist, entrepreneur and author. He is the Founding Partner of Golden Gate Ventures, an international venture capital firm investing in high-growth markets across Asia and the Middle East. Founded in 2011 and headquartered in Singapore, the firm manages more than US$300 million and has invested in over 150 companies, including nine unicorns and two IPOs. Vinnie has served on the boards of more than a dozen startups, including unicorns Carro and Carousell. He was also an early investor in Stripe and AngelList and helped companies such as Lomotif navigate their path to the public markets.
Before founding Golden Gate Ventures, Vinnie built two startups in Silicon Valley. The first, Meetro, was a pioneering location-based chat application that ultimately closed but provided many of his earliest lessons as an entrepreneur. He later founded Lefora, a social-network platform that was acquired by a public media company in 2010. He also founded Silicon Valley NewTech, which became one of the Bay Area’s largest startup demo events at the time.
Vinnie is the co-author of Mind the Gap: Scaling Businesses Across Cultures, a Straits Times bestseller and #1 Amazon bestseller in Business Entrepreneurship. Drawing on the experiences of more than 50 CEOs and business leaders, the book explores the cultural dynamics that can determine whether international expansion succeeds or fails.
He has been a guest speaker and lecturer at Stanford University, the University of California, Berkeley, Carnegie Mellon University in Qatar and the National University of Singapore, and is a Kauffman Fellow. He holds a degree in Computer Engineering from Boston University and later pursued an EMBA at Tsinghua University, reflecting his long-standing interest in bridging Silicon Valley and Asia.
Earlier in his career, Vinnie spent four years at IBM, helping shape the company’s strategy around social software for the enterprise. A lifelong technologist who joined his first startup as a programmer while still in high school, he is passionate about helping founders build internationally successful companies and strengthening startup ecosystems around the world.