Amex AI Turns Slower Hiring Into Quiet Headcount Cuts

If Amex AI lets American Express add fewer service representatives without announcing mass layoffs, when does productivity become a quiet headcount cut?

That is the real labor story inside Amex’s latest earnings. CEO Steve Squeri said the company is using AI to move faster through technology work and is not adding customer service representatives at the old pace, according to American Banker. My view: Amex is showing how white-collar automation will reshape payroll first through slower hiring, attrition and higher output targets, not through one dramatic layoff memo.

"It's not even early innings yet, it's the preseason," Squeri said.

That line was meant to signal caution. It should also make workers pay attention.

Is Amex AI already changing headcount before layoffs become the headline?

Yes. The key phrase from Amex was a "lack of acceleration in hiring" in customer service, even as business volume for that group increased. Squeri was blunt about the direction.

"The number of reps hasn't grown. Over time it will decrease from attrition," Squeri said.

That is not a mass firing. It is still a labor shift.

Customer service is one of the first places AI can bite because the work has repeat questions, searchable policies, standard workflows and measurable response times. Amex this year launched an AI-powered service portal for service representatives designed to reduce call time. If that tool helps one representative handle more volume, management does not need to replace every departing worker.

This is how automation usually becomes durable. A company does not need to fire 5,000 people to change the shape of its workforce. It can hire 500 fewer people over time. Every unfilled role compounds.

Amex also has a constraint that cheaper brands may not face as sharply: service quality is part of the product. A premium card company cannot let AI turn customer support into a maze of canned responses. If Amex AI reduces friction while preserving accuracy, it strengthens the franchise. If it cuts too hard, it risks damaging the thing cardmembers actually pay for.

Does the technology backlog matter more than the customer service slowdown?

Probably. The customer service signal is easier to grasp. The more important comment may be Squeri’s point about technology work.

"We have a large backlog of tech products, with AI we are getting to more things quicker," Squeri said.

That should land with anyone who has worked inside a large financial firm. Backlogs are hidden costs. They delay product launches, slow internal fixes, leave data work unfinished and force teams to keep patching around old constraints.

Amex did not give a detailed breakdown of which tools are clearing that backlog. So we should not pretend to know more than the company disclosed. But the business logic is clear: if AI can help teams draft code, test changes, summarize documentation, route work and automate pieces of internal processes, the first payoff is not a flashy new consumer product. It is speed.

That matters because Amex’s earnings already gave investors a strong base. For the second quarter, the company reported net income of $3.11 billion, up from $2.88 billion a year earlier. Earnings per share rose to $4.53 from $4.08. Revenue climbed to $19.64 billion from $17.8 billion. Billed business reached $445.8 billion, up from $416.3 billion.

Analysts expected quarterly earnings of $4.40 and revenue of $19.62 billion, according to Zacks Investment Research, as cited by American Banker. Amex also projected full-year growth of 10%, a slight increase from its earlier 9% to 10% range, while affirming profit guidance of $17.30 to $17.90 per share.

When a company with those numbers says AI is helping it move faster, investors will hear margin expansion. Workers will hear a different question: faster for whom?

Are shareholders getting the cleanest version of the Amex AI story?

They are. The shareholder case is neat.

If American Express can handle more customer service volume without growing the rep base, and if it can clear technology work faster without matching every project to more staffing, operating leverage improves. Earnings growth becomes less dependent on higher spending, richer fees or endless new card incentives.

That is the appeal of Amex AI from Wall Street’s seat. Productivity drops into the model before the social consequences show up in the org chart.

The labor issue is messier. Entry-level service jobs can be more than cost centers. Inside large financial companies, they can be training grounds. Workers learn products, customer behavior, compliance expectations and internal systems. Some move into better roles later.

If AI quietly reduces the number of those jobs, the company may save money while narrowing a path into the business. That is not an argument against AI. It is an argument against pretending attrition has no human cost because it does not arrive as a pink slip.

There is a better way to frame this. If Amex is gaining productivity from workers and AI systems operating together, some of that gain should flow back into training, mobility and better jobs. Otherwise, the bargain becomes one-sided: employees absorb the pressure, shareholders collect the upside.

For the payments side of this debate, XOOMAR has tracked how controls and automation are reshaping finance operations in Mastercard Virtual Cards Lock Down B2B Spend Controls. The same principle applies here. Better systems can reduce waste. They can also harden into tighter control over labor.

Could better AI tools actually make Amex jobs better?

Yes, and this is the strongest defense of Amex’s strategy.

AI can strip out tedious work. It can reduce call handling friction. It can help service reps find the right answer faster. It can rescue technology teams from backlog purgatory. Slower hiring is not the same as firing workers to flatter a quarterly expense line.

The evidence from inside Amex supports that defense. VentureBeat reported that Amex’s internal IT chatbot has produced a 40% increase in its ability to resolve IT queries without transfer to a live engineer, after the company infused generative AI into the tool. The same report said a dedicated council initially identified 500 potential use cases, narrowed to 70 in various stages of implementation.

That is not trivial. It suggests Amex is treating AI as infrastructure, not a toy.

VentureBeat also reported that Amex’s travel counselor assist tool supports 5,000 travel counselors across 19 markets, with more than 85% saying it saves time and improves recommendation quality. Hilary Packer, Amex EVP and CTO, framed the internal support gain plainly:

“It's giving people the answers, as opposed to a list of links,” Packer told VentureBeat. “Productivity is improving because we're getting back to work quickly.”

That is the version of AI workers can accept: fewer dead ends, better information, less repetitive searching.

The harder part is governance. If AI makes every rep faster, does Amex reduce pressure or raise targets? If technology teams clear more work, do they get breathing room or a bigger queue? The difference determines whether AI improves jobs or simply turns every role into a higher-speed version of itself.

Training will decide much of that. So will transparency. XOOMAR’s coverage of Synthesia AI Roleplay Sessions Grill Workers Live at Work shows how AI is already entering workplace preparation and evaluation. Companies using these tools need to be explicit about whether they are coaching people or just measuring them harder.

Should Amex turn productivity gains into a worker compact instead of a silent hiring squeeze?

It should. And so should every financial company watching Amex.

The practical steps are not mysterious.

• Disclosure: Explain how AI is changing hiring, roles, training and promotion paths.
• Retraining: Move customer service staff toward AI-supervised work, exception handling and higher-value support.
• Mobility: Create internal pathways from service operations into product, risk, data and technology roles.
• Quality controls: Measure service accuracy and customer outcomes alongside cost savings.
• Worker voice: Let employees report where AI helps, where it fails and where it adds pressure.

Amex has another AI frontier to manage: agentic commerce. Squeri said that with agentic commerce, “it is fraught with not only fraud but also hallucinations,” while arguing that Amex’s closed loop model gives it data from both the customer and merchant sides. The company also supports the Agent Payments Protocol (AP2), an open standard launched in late 2025 alongside Google, PayPal and more than 60 other companies.

That raises the stakes. If AI agents can discover products, make decisions and complete payments, trust becomes even more valuable. Financial companies do not sell software alone. They sell confidence.

The next few quarters will not answer the whole question. Squeri is right that it is early. But Amex has already shown enough to make one point clear: the companies that win with AI will not be the ones that simply add fewer people. They will be the ones that make the people they keep more valuable.

Disclaimer: This XOOMAR analysis is for informational and educational purposes only. It is not financial, investment, legal, tax, or professional advice. It does not provide buy, sell, hold, price-target, portfolio, or personalized recommendations. Verify information independently and consult qualified professionals before making decisions.

Impact Analysis

• Amex shows how AI can reduce workforce growth without headline layoffs.
• Customer service roles are especially exposed because AI can speed up repeatable workflows.
• Workers may face higher productivity expectations as companies use AI to handle more volume with fewer hires.