For years, foreign exchange and cross border remittances were treated as a specialist corner of banking. Customers walked into a branch, called a dealer or used a basic online form. Today, much of that activity has moved to apps and web portals. AI powered chatbots sit on the surface, answering questions about rates and limits at any time of day.
Yet, marketers inside banks and fintech firms say the first wave of chatbots is no longer enough. Volatile currencies, tighter regulations and demanding customers are forcing a shift from simple conversational tools to what many now call smart doers, systems that can understand intent and also take action inside core processes. For FX marketers, the real opportunity lies in designing these smart doers so that every interaction supports acquisition, cross sell and loyalty, not just query resolution.
The context in India underlines why this matters. The country has been the world’s largest recipient of remittances in recent years, with inflows crossing 120 billion dollars and touching millions of households. At the same time, a global survey by a large technology firm’s business value institute reported that about 90 percent of CEOs expect their companies to adopt generative AI across business processes and that more than 80 percent of senior executives believe such systems will interact directly with customers in the near term. In other words, the FX business is both high stakes and ripe for automation.
Banking leaders in India already see AI as a major lever in this landscape. At a recent financial services roundtable, Rohit Raina, Vice President and Vertical Head, Direct Marketing at HDFC Bank, summed up the shift simply. “Marketing will be the biggest beneficiary of AI. It has huge relevance for hyper personalisation. It can understand the customer’s persona or you can offer persona based prompts,” he said. That logic applies strongly to FX and remittances, where timing, context and trust define whether a customer completes a transaction or waits.
Chatbots as the first line, not the finish line
The current generation of chatbots already plays an important role in Indian banking. Many large institutions route basic queries through bots on their websites, apps and messaging channels. These systems handle routine questions about exchange rates, documentation, cut off times and status checks.
Ankit Goenka, Head of Customer Experience at Bajaj Allianz General Insurance, offered a clear view of how far such tools have come. “Chatbots are the success story of AI. But affordability is key,” he said at the same roundtable, noting that even farmers are now comfortable talking to a bot in local languages. For frontline staff, that first layer of automation removes repetitive work and frees time for more complex cases.
In another example, Aditya Kulkarni, Deputy Vice President and Head, Business Intelligence Unit at Star Union Dai ichi Life Insurance, explained how their organisation treats bots as a core part of the customer front door. “We have seamlessly integrated bots as the first line of customer interaction,” he said, describing how queries are triaged before being passed on to human agents where necessary.
FX and remittance teams have adopted similar approaches. Bots answer questions such as “what is today’s dollar rate”, “how much can I send to Canada for education” or “why is my transfer pending”. That layer is now considered basic hygiene. The emerging question for marketers is different. If a bot can hold a conversation, can it also nudge the customer to the right action at the right time, and can it do so while respecting regulatory and risk constraints in FX.
From answering to acting
This is where the idea of a smart doer comes in. A smart doer is not a new channel. It is an AI powered agent that can interpret free form queries and then initiate steps inside back end systems within agreed limits. For example, when a customer types, “my daughter’s fees are due next week in Canada, should I lock the rate now or wait”, a traditional bot might show a generic answer on rate trends. A smart doer goes further. It checks the customer profile, previous transfers, current rate alerts and any consent already given, then offers concrete choices such as creating a rate alert, starting a pre filled transfer request or connecting to a dealer for a larger amount.
In a corporate FX context, a mid sized exporter might ask, “show me my dollar exposures for the next three months and suggest simple hedges”. A smart doer can pull data from the treasury management system, present a simple view and tee up standard hedging options for approval. The marketer’s role here is not to hard sell products inside the bot, but to ensure that the journeys and scripts are designed around meaningful outcomes such as reduced leakage to competitors, higher stickiness and smoother experiences.
Several Indian and global banks are already experimenting with such agent style systems in pilots. Internally, they often sit as AI copilots for relationship managers, pulling together client history, deal pipelines and rate views into a single screen. The same logic can extend to customer facing tools once risk teams are comfortable with guardrails.
Designing journeys for FX customers
For FX marketers, the shift from chatbot to smart doer changes how journeys are mapped. Instead of a static FAQ tree, teams now chart flows around specific jobs to be done. Common FX journeys include sending money abroad for education, supporting family, paying suppliers, receiving export proceeds or managing travel expenses.
In each case, marketers can define a few canonical paths. For example, an education remittance journey might start months before admission, with content that helps families understand limits, documentation and timelines. As offers and visas arrive, the smart doer can recognise changes in behaviour, prompt pre funding of accounts or suggest scheduling transfers before expected rate events. After the first transaction, it can automate reminders for semester fees, always within the rules of the central bank and the bank’s own risk appetite.
These flows need deep integration. A smart doer must be able to read data from CRM systems, core banking, transaction monitoring tools and marketing automation platforms. It has to respect flags set by compliance and risk, for instance when a transfer needs additional checks. It also needs the ability to pause, route to a human or ask for explicit confirmation when it reaches the boundary of what it is allowed to do alone.
Why marketers care about orchestration
From a marketing standpoint, the promise of such systems goes beyond better service. FX is a high intent, high involvement category. When customers are in the market, they actively compare providers on rate, speed, documentation and trust.
AI agents give marketers a way to show up contextually without adding friction. For instance, instead of sending generic email campaigns about remittance offers, teams can trigger targeted prompts only when a customer shows intent signals such as searching for rate tables, logging into education sections or checking balances in foreign currency accounts. The smart doer then becomes the surface where those prompts are delivered in natural language.
Rohit Raina’s point about AI enabling persona based prompts is relevant here. In practice, this means the system recognises whether the customer behaves like a frequent remitter, an occasional traveller or a first time sender. Each persona receives different nudges, tone and level of detail, even when asking similar questions about rates or limits.
For this to work, marketing, sales and product teams must agree on a shared view of the customer. That includes standard definitions for value segments, risk banding and eligibility for offers. It also includes a feedback loop where outcomes from smart doer interactions such as completed transfers, abandoned flows or escalations feed back into audience models.
Limits, risks and the human in the loop
Despite the enthusiasm around AI, Indian financial institutions remain cautious, particularly in regulated categories like FX. Leaders consistently highlight the need for responsible use, clear guardrails and human oversight. Banks have learned from early chatbot deployments that customers will test systems with unexpected queries, emotional language and edge cases such as suspected fraud.
This is why most serious FX automation initiatives still keep humans in the loop for high value or sensitive actions. A smart doer might pre fill details, suggest actions and collect consent, but a human dealer or service agent often approves or supervises the final step for large transactions. Some organisations also maintain clear thresholds where any deviation from normal patterns, such as unusual counterparties or sudden spikes in amount, automatically triggers manual review.
Affordability, which Ankit Goenka flagged as critical, remains a structural constraint. Advanced AI models and tight integrations are expensive to deploy and maintain. However, providers report that the unit economics improve once volumes scale and once the same orchestration layer supports multiple journeys across products, not just FX. For smaller institutions, cloud based platforms and shared infrastructure are making it more practical to experiment.
What smart doers mean for teams
The rise of smart doers does not eliminate the need for human marketers. Instead, it changes what they spend time on. Rather than writing one off campaign copy or managing long approval chains, more time goes into designing reusable journeys, specifying guardrails and reviewing data. There is also growing demand for hybrid profiles who understand both FX products and AI tooling, including prompt design, conversation flows and basic analytics.
Within FX desks, relationship managers are beginning to see AI agents as copilots that summarise client positions, surface opportunities and handle low complexity follow ups. In marketing teams, AI tools draft variants of educational content, FAQs and alerts, which humans then refine for accuracy and local nuance.
The broader employment picture in BFSI supports this direction. Industry leaders repeatedly stress that AI will augment rather than simply replace roles, especially where regulation, judgment and trust are involved. Systems can handle volume and pattern recognition. Humans still frame strategy, interpret exceptions and own the relationship.
Where FX marketers can start
For FX marketers looking to move beyond chatbots, industry practitioners suggest a few practical steps. The first is to map the one or two journeys where delays or friction most often cause drop offs, for example first time education transfers or repeat small ticket remittances. These are strong candidates for smart doer pilots.
Next, teams can define clear actions that the agent is allowed to take autonomously, such as setting rate alerts, collecting and validating basic documentation, or scheduling a call back within promised timelines. Everything else routes to a human. This scoped approach keeps risk manageable.
Third, marketers should partner closely with technology, legal and compliance functions to ensure that any use of AI agents respects regulatory guidance, particularly around know your customer norms, data privacy and consumer communication. Transparency matters. Customers should know when they are interacting with an AI system and when they are speaking to a human.
Finally, measurement needs to focus on meaningful outcomes. Instead of only counting chat volume, teams can track how smart doer interactions influence completed transfers, repeat usage, cross sell and satisfaction scores. Over time, these metrics help build the business case for broader AI investments in FX.
The emerging consensus from Indian banking and insurance leaders is that chatbots have earned their place as the first layer of interaction. As Aditya Kulkarni described, they can be seamlessly integrated into the front door of customer engagement. The next phase for FX marketers will be shaped by smart doers that not only talk, but also help customers act with confidence inside complex, regulated journeys. The challenge is to make those systems reliable, affordable and firmly aligned with human judgment, so that automation strengthens rather than weakens the trust that underpins every cross border transaction.
Disclaimer: All data points and statistics are attributed to published research studies and verified market research. All quotes are either sourced directly or attributed to public statements.