Bank of America Deploys Generative AI to 18,000 Contact Centre Agents

What happened

Bank of America has embedded generative AI capabilities into EricaAssist, its internal agent-support platform, making the technology available to approximately 18,000 contact centre agents. Rather than deploying a customer-facing chatbot, the bank has chosen to augment its human workforce — giving agents real-time access to client insights and contextual information during live interactions.

The move represents a deliberate architectural choice: generative AI sits behind the agent, not in front of the customer. EricaAssist already underpinned the bank's contact centre operations; the generative layer adds the ability to surface relevant account context, suggested responses and supporting information at the moment an agent needs it most.

Why it matters

The decision to prioritise agent augmentation over direct customer-facing automation is significant for anyone designing service experiences at scale. There is a well-documented tension in contact centre AI deployments between cost-reduction ambitions — which push towards self-service automation — and quality outcomes, which often depend on human judgement, empathy and contextual reading of a customer's emotional state. Bank of America's approach signals a belief that the highest near-term value from generative AI lies in compressing the time it takes a human agent to reach confident, accurate answers, rather than in removing the human altogether.

From a behavioural economics perspective, this matters because customer trust in financial services is fragile and asymmetric: a single poor interaction can erode loyalty that took years to build. Equipping agents with richer, faster context reduces the cognitive load on the frontline and lowers the probability of the hesitations, transfers and errors that damage perceived competence. For CX and service-design practitioners, it is a reminder that "AI in the contact centre" is not a single design decision — where in the service journey the AI sits changes the experience outcome entirely.

By the numbers

• 18,000 contact centre agents now equipped with generative AI-powered insights through EricaAssist.

The Renascence take

Most commentary on AI in financial services fixates on deflection rates and cost-per-contact. Bank of America's deployment invites a more nuanced question: what does AI do to the quality of the human moment that remains?

The instinct to automate the customer interaction is understandable but often premature — particularly in high-stakes, high-emotion categories like personal finance. Placing generative AI in the hands of the agent rather than in front of the customer is a behaviorally sounder design: it preserves the relational signal customers value while reducing the informational bottleneck that makes agents hesitate or escalate unnecessarily. What most operators will miss is that this is not a conservative halfway house — it is a deliberate sequencing strategy. The organisations that will win are those that use this phase to measure how AI-assisted agents perform on resolution quality and emotional attunement, not just handle time, and then let that evidence guide how far automation eventually extends toward the customer.