Why Better Banking Depends More on Better Timing Than Better Data
By Jessica Kendall, Contributor at The Financial Brand
Simple Subscribe
Subscribe Now!
After more than 3.2 billion customer conversations through Erica, Bank of America has learned something surprising: consumer banking questions are remarkably predictable, but customer needs are deeply emotional. In a recent conversation on the Banking Transformed podcast, Jorge Camargo, Head of Digital Platforms at Bank of America and former Walt Disney Company executive, shared how Erica evolved from a digital assistant answering basic questions into enterprise infrastructure powering experiences across consumer banking, wealth management, employee support and treasury services.
Key takeaway: Camargo argues that AI in banking is moving through three stages — answers, actions, and autonomy — but success depends less on technology than operational discipline. The biggest opportunity for banks is using AI to help customers answer a broader question: “Am I okay?”
In this world, trust, human support, and organizational alignment matter as much as model sophistication.
Need to Know:
- Personalization is a timing problem, not a data problem. Relevance depends on surfacing the right information at the right moment rather than overwhelming customers with insights.
- Most customer interactions are low stakes, but critical moments define relationships. Banks need systems that recognize moments of stress, urgency, or financial significance and respond accordingly.
- AI adoption requires process redesign, not just technology investment. Institutions that rethink workflows around AI stand to benefit more than those layering tools onto existing systems.
- Trust grows through proactive usefulness and clear human off-ramps. Customers are more likely to engage with AI when institutions make escalation easy and demonstrate restraint.
- Smaller institutions may have an advantage. Without legacy complexity, regional and community banks can move faster by redesigning processes around modern AI capabilities.
What Billions of Conversations Reveal
When Bank of America launched Erica nearly a decade ago, the goal was to help customers find answers faster. Over time, however, the bank discovered that digital conversations contained something more valuable than efficiency gains. They offered clarity about what customers care about.
After processing 3.2 billion interactions, Camargo said Bank of America found that consumer banking could largely be distilled into roughly 700 recurring topics. That realization transformed Erica from a chatbot into something far more strategic: a system capable of identifying patterns, anticipating needs and improving experiences across the organization.
Yet the most important lesson had little to do with automation. Customers rarely come to their bank with isolated questions. Instead, many are trying to answer a broader concern: Am I good?
That question surfaces whether someone is paying down debt, managing investments. or navigating everyday finances. A balance inquiry may really be about financial anxiety. A declined transaction might create immediate stress. A transfer request may signal a major life milestone.
What this means: The insight fundamentally changed how Erica operates. Today, roughly 60% of Erica’s interactions are proactive rather than reactive. Instead of simply responding to questions, the platform increasingly surfaces information that customers may need before they ask.
This points to a meaningful shift in thinking. AI’s value increasingly lies in reducing uncertainty and helping customers feel more informed and in control. This can mean:
- Flagging unusual spending patterns before they become a problem
- Predicting future balances based on recurring transactions
- Reminding customers about permissions granted to third-party financial apps
- Offering timely guidance during periods of financial stress
The real advantage comes from moving beyond transactional service toward contextual guidance.
Why Timing Matters More Than Data
Banks have spent years talking about personalization while customers still receive generic product offers that miss the moment entirely. Camargo frames personalization differently. The challenge is not how much information a bank has — most institutions already possess enormous amounts of customer data.
The challenge is timing. When should the institution step in with something genuinely useful? He shared a practical example involving declined card transactions. By integrating data from card systems into Erica, the platform can identify a failed payment and proactively ask whether the customer needs help before frustration escalates. That responsiveness matters because customers increasingly compare banking experiences against digital leaders in other industries.
Camargo’s years at Disney shaped much of this thinking. Disney became famous for experiences that felt magical while remaining almost invisible operationally. The lesson he brought into banking is that meaningful personalization often comes from small moments that demonstrate awareness without crossing into discomfort.
Key insight: Banks frequently struggle with the line between helpful and invasive. Customers appreciate relevance, but excessive targeting can feel unsettling. To get it right, it comes down to intent and execution. Helpful personalization answers a real need. Intrusive personalization feels performative.
Banking executives should prioritize experiences that reduce friction, improve confidence, or solve problems over those designed purely to market products.
Why Trust Still Requires Humans
Despite rapid advances in AI, Camargo repeatedly emphasized that human support remains essential. At Bank of America, one of Erica’s most important capabilities is recognizing when to step aside.
Customers can escalate to phone support, live chat, or in-person appointments, with Erica carrying forward context to avoid forcing customers to repeat themselves. That continuity matters during high-stakes situations — the moments that disproportionately shape customer trust.
Camargo estimates that roughly 95% of banking interactions are relatively routine: checking balances, paying bills or moving money. But the remaining moments carry outsized emotional weight, such as dealing with a death in the family, sending a down payment for a home, responding to fraud concerns, or navigating financial hardship after disasters.
In these moments, speed and empathy matter more than digital efficiency.
Key insight: This philosophy also shapes how Bank of America innovates internally. Rather than treating legal, compliance, and risk teams as final approvers, Camargo said they were brought into the design process from the beginning. That approach helped support more than 85,000 updates to Erica since launch — from small refinements to major capabilities — without creating bottlenecks.
Stop Chasing Technology Alone
Camargo offered particularly useful advice for smaller institutions trying to catch up. His recommendation was don’t copy Erica. Instead, banks should copy the discipline that built it.
Technology itself has become increasingly accessible. In many cases, regional and community institutions can deploy capabilities faster than large enterprises burdened by complexity. The differentiator will not be who has access to AI, but who redesigns operations around it.
Camargo recommends a simple exercise: imagine an AI advanced enough to do almost anything. Then divide work into two categories — what should remain uniquely human and what technology should handle. From there, banks can begin redesigning workflows accordingly.
Bottom line: The institutions that gain the most from AI may not be those with the biggest budgets. They may be the ones willing to rethink business processes from the ground up. Bank of America’s experience suggests the future of banking AI is not simply smarter chatbots. It is creating systems that anticipate needs, reduce friction, and free customers to focus less on managing money — and more on using it to build the lives they want.
