Bank of America upgraded EricaAssist on July 21, 2026. The tool sits as a small widget on the screen of a customer service employee while they are on a call. It reads the conversation, identifies why the customer is calling, checks the customer's history with the bank, and suggests the next best action, all within three seconds.
The employee stays in control. The customer does not know the tool exists. The AI does not talk to the customer directly.
The result so far: call times are down by close to one minute per interaction. That sounds modest, but across 18,000 employees handling millions of calls, it adds up to a significant reduction in cost and a faster experience for customers on the other end of the line.
This is the direction all large customer service operations are heading. Bank of America has been building this capability since 2018, when it launched Erica, the customer-facing assistant that now handles over 58 million interactions per month. EricaAssist is the internal version, built for staff rather than customers. The bank also has an internal general-purpose AI assistant used by over 90% of its 213,000 employees, which has cut internal IT support calls by more than 50%.
The spending behind this is serious. Bank of America puts $14 billion a year into technology overall, with more than $4 billion of that going toward new initiatives including AI. This is not experimental spending. It is operational investment at scale.
The rest of the banking industry is moving in the same direction. JPMorgan says its $2 billion annual AI spend has produced comparable savings. KeyBank reported that AI-handled calls cost around 35 cents each, versus $9 for a human-handled call. Citi has made AI prompt training mandatory for most of its staff. Across the industry, 47% of US banks had already deployed generative AI as of a 2025 survey, up from just 10% in 2023.
The workforce picture is more nuanced than the headlines suggest. A 2026 American Banker survey of 206 bank executives found that only 3% said AI had led to workforce reductions at their firm so far. The bigger trend at the moment is augmentation: staff doing the same job faster and with fewer errors, rather than being replaced outright. Banks are actually hiring more AI-proficient staff, with AI-related job postings at US banks up over 77% in 2025.
But the longer-term picture is different. Morgan Stanley analyzed 35 European banks and projected that around 10% of banking jobs, roughly 212,000 roles, could disappear by 2030, concentrated in back-office and middle-office functions where work is repetitive. Wells Fargo's CEO has said publicly that AI will have a significant impact on headcount over time.
For any business operator outside of banking, the pattern here is the template. AI is not replacing front-line staff in one move. It is being inserted as a real-time assistant that makes each employee more capable. The immediate effect is speed and consistency. The longer-term effect is that you need fewer people to handle the same volume, or the same number of people handling much more volume.
If you run any kind of customer service team, in insurance, retail, logistics, or elsewhere, this is the model to watch. The tools exist. The cost is coming down. And your competitors in every industry are paying close attention to what the banks are proving out at scale.