- Workforce Cuts: Uber, Microsoft and Commonwealth Bank reportedly cut customer-service staffing while expanding artificial intelligence automation.
- Task Automation: AI agents handle repetitive requests around the clock, while people retain sensitive, technical and judgment-heavy cases.
- Evidence Limit: Many staffing totals remain unconfirmed by the companies and rely on one originating report.
- Worker Impact: Outsourcing hubs face pressure, while roles may emerge in AI supervision, escalations, data curation and prompt engineering.
Uber’s expansion of artificial intelligence across its operations included a reported cut of about 10 percent of its customer-service workforce in July. Routine support work, from basic requests to common account problems, sits closest to the reduction.
Kate Leggett, a Forrester analyst, estimated that “almost half of customer service roles will be impacted by 2030”. Leggett’s forecast measures the exposure: affected roles can lose tasks, change duties or be eliminated as software handles more interactions.
AI agents are increasingly able to complete complex multi-step support tasks rather than merely suggesting replies. Companies like Uber can assign them routine contact-center tasks around the clock. However, sensitive, technical and judgment-heavy cases still need to reach specialists when customers request human help.
Cost Savings Become Head-Count Pressure
Workforce reductions in the tech-support sector are ongoing. Microsoft’s support workforce reportedly fell from about 50,000 to 40,000 employees and contractors while AI was saving about $750 million annually in support costs as of April 2026. A person familiar with Microsoft’s operations shared the workforce figures with Bloomberg. Sales and service operations chief Judson Althoff gave the savings figure as the company expanded Dynamics 365 service automation.
Commonwealth Bank of Australia shed hundreds of chat-support workers while bank officials said it added more than 140 Australia-based call-center roles over the preceding six months.
Brinks Home Chief Information Officer Philip Kolterman said AI cut call volume by about two-thirds as its call-center workforce declined from roughly 800 to 400. Some displaced workers moved to other divisions, while attrition accounted for the majority of remaining reductions. Across the three companies, domestic hiring, transfers and unfilled departures softened some immediate dismissals without preserving the same jobs, employers or locations.
Routine Work Moves to Agents, Not Every Human Role
Microsoft’s own AI products have moved beyond demonstrations. Service Agent in Microsoft 365 Copilot is generally available since June 30. Autonomous email resolution remains planned for September 2026, preserving the distinction between a deployed support agent and a future capability.
Earlier deployments brought autonomous customer-service agents to market in October 2024, generative AI features for routine contact-center tasks in December 2024 and customer-service AI agent builders in March 2026. Acting on cases, accounts and emails gives the software a route to remove more tasks from frontline queues without treating every customer conversation alike.
Employers can automate repetitive interactions while retaining people for escalations, disputed payments and conversations requiring judgment or empathy.
1-800Accountant expects to spend 50 percent less on outsourced staff in 2027 after automating simple tasks with Salesforce’s Agentforce platform. Ryan Teeples, the company’s chief strategy officer, framed the expected savings as a risk for major outsourcing hubs:
“I wouldn’t want to be an investor in India or the Philippines right now, because that’s where we are seeing the biggest cost savings from AI.”
Ryan Teeples, 1-800Accountant chief strategy officer (via Bloomberg)
India and the Philippines face particular exposure because business process outsourcing (BPO), or contracted support work, concentrates large teams around repetitive interactions. Software can reduce billable volume by absorbing the routine tier while people handle exceptions. Geography and task complexity shape which workers encounter pressure first.
Some work may shift toward AI supervision and escalated disputes, training-data curation and prompt engineering. Triveni Singh, a cybercrime expert and former Indian Police Service officer, urged support professionals to develop data-management, AI-tool and advanced problem-solving skills, though upskilling offers no employment guarantee.
Microsoft’s plan to resolve customer emails autonomously with AI targets general availability in September 2026. If the schedule holds, another routine support channel would become available for customer deployment.
