Where the Customer Service Job Is Actually Headed
4 min read
This is a new occasional format for the site: instead of covering what's already happened, we're using real, current data to make a hedged prediction about where something is headed next. First up is a narrow one — not "AI and jobs" broadly, but the specific fate of the customer service call center, and the millions of people, in the US and especially overseas, whose paycheck currently depends on it.
The trend line so far
Gartner, the research firm most enterprises use to plan technology budgets, predicted in March 2025 that by 2029, "agentic AI" will autonomously resolve 80% of common customer service issues without a human involved, cutting operational costs by 30%. Worth naming the obvious caveat: Gartner sells advisory services to the same companies making these buying decisions, so treat it as industry-informed forecast, not neutral fact. But the number lines up with 2026 hiring data: Salesforce, Verizon, Oracle, Klarna, and Monday.com have all cited AI adoption as a factor in contact-center workforce reductions this year, contributing to a reported 100,000-plus tech-sector job cuts tied to AI transformation.
The exposure isn't evenly spread. It concentrates hardest in the business-process-outsourcing (BPO) industry — companies answering customer service calls for others, largely from the Philippines and India. The Philippine BPO sector alone directly employs around 1.8 to 1.9 million people and generates roughly 7 to 8% of the country's GDP, with close to 70% of that revenue still tied to voice-based support — exactly the routine, scriptable work current AI handles best. Estimates of what's at risk vary widely, from the IMF's more conservative ~3% of the Philippine workforce in immediately displaceable roles, up to the Philippine Department of Trade and Industry's own estimate that half of all BPO jobs face some displacement risk by 2030. Even the low end of that range is tens of thousands of real households.
The part that complicates the story
If the trend stopped there, this would just be a doom post, and that's not what the evidence supports. The clearest counter-data point is Klarna, which became the industry's cautionary tale in real time. In 2024 the company replaced the equivalent of 700 support agents with an AI assistant, touting roughly $40 million a year in savings. By May 2025, CEO Sebastian Siemiatkowski was publicly walking it back, telling Bloomberg that cost had been "a too predominant evaluation factor" and quality had suffered — customers wanted a human for anything that wasn't routine. Klarna spent the following year rebuilding a hybrid model: AI handles high-volume routine queries, humans handle escalations and high-value accounts. Not a story about AI failing to work — a story about where the current ceiling actually sits, the same shape showing up industry-wide.
The hedged prediction
If this trajectory continues, the most likely outcome over the next three to five years probably isn't "AI eliminates the call center job" — that's the sci-fi version this format is explicitly trying to avoid. It's a bifurcation. The routine, high-volume, low-complexity layer of customer service — password resets, order status, simple returns — keeps shifting to AI at a real and possibly accelerating pace, likely closer to Gartner's 2029 timeline than to a slower rollout, given how much cost pressure is already visible in 2026 earnings calls. What's left is a smaller, more specialized layer: escalation handling, emotionally sensitive cases, high-value accounts, and a new tier of jobs overseeing and correcting the AI layer itself — quality review, conversation design, the people who get called in when the bot gets it wrong.
The cost of that split falls unevenly, and it's worth naming plainly rather than softening it. It falls hardest on entry-level and routine-tier workers, and disproportionately on BPO-dependent economies that built regional employment around exactly the layer of work AI now does most cheaply — with limited local alternative industries to absorb the loss, unlike a Verizon or Salesforce employee in a larger, more diversified US labor market. The benefit — faster, cheaper, always-available support, and a smaller pool of better-paid, more specialized human roles for the workers who make the transition — is real too, but it isn't the same population capturing both sides of that trade. If you or someone you know works in this field, that mismatch, not the abstract "AI will change jobs" headline, is the part actually worth planning around.
We'll revisit this one directly if the numbers move meaningfully before 2029 — that's the point of writing the prediction down now.