The contact center industry employs roughly 17 million people globally. It’s also the industry that AI is disrupting most aggressively. The economics are brutal for human agents and irresistible for businesses: an AI agent costs roughly $0.50-2.00 per resolved conversation, compared to $5-15 for a human agent. It works 24/7, speaks dozens of languages fluently, never gets frustrated, and improves with every interaction. The result is the fastest labor transformation since manufacturing automation.
The technology leap
Customer support AI has evolved through three generations in rapid succession. First-generation chatbots (2016-2020) used decision trees and keyword matching; they could handle “what’s my order status?” but collapsed at the first unexpected question. Second-generation systems (2021-2023) used fine-tuned language models that handled more variety but still required extensive training on company-specific data and frequently hallucinated policies or prices.
Third-generation systems (2024-2026) are built on frontier models with retrieval-augmented generation (RAG), function calling, and persistent memory. They can access customer history, CRM records, knowledge bases, and order systems in real time. They can take actions — issue refunds, modify subscriptions, schedule appointments — not just provide information. And they maintain context across channels: a conversation that starts on web chat continues seamlessly on the phone, with the AI remembering everything.
Intercom’s Fin AI exemplifies the current state of the art. It resolves roughly 65% of customer inquiries without human intervention, with a customer satisfaction score within 3 points of human agents. Its “AI-first” routing means customers interact with AI by default, with human escalation only when the AI reaches its confidence threshold or the customer explicitly requests it.
Zendesk AI has taken a different approach, focusing on agent augmentation rather than replacement. Its AI suggests responses, retrieves relevant knowledge base articles, and automates post-interaction summarization, but keeps humans in the driver’s seat for actual customer communication. This “copilot” model has been more popular with enterprise customers who are cautious about fully autonomous customer-facing AI.
The labor impact
The numbers are stark. A 2026 Gartner analysis projects that AI will handle 45% of all customer service interactions by 2028, up from roughly 15% in 2025. For routine inquiries — password resets, order tracking, basic troubleshooting — the figure is already above 50% at many organizations.
The impact on employment is uneven. Entry-level support roles are declining rapidly. But demand for “AI conversation designers” — people who craft the personality, tone, and escalation logic for AI agents — has exploded, with salaries rivaling software engineers. The new customer support career path looks less like answering phones and more like managing a fleet of AI agents: monitoring performance, handling edge cases, and continuously improving the system.
Several large employers have committed to retraining programs rather than layoffs, transitioning front-line agents into AI supervision, quality assurance, and complex case handling roles. But the net effect on employment is almost certainly negative — AI handles more conversations with fewer total human hours.
Where humans still win
For all the progress, there are scenarios where human agents remain essential — and likely will for years.
Emotionally charged situations — a bereaved family member closing a loved one’s account, a small business owner facing bankruptcy due to a service outage — require genuine human empathy that AI convincingly simulates but doesn’t actually possess. Most companies wisely keep humans in the loop for these high-stakes interactions.
Complex, multi-system problem solving also remains a challenge. When a customer’s issue spans billing, technical, and account management systems with inconsistent data, human agents can navigate the ambiguity and advocate internally in ways AI cannot.
Trust is another factor. A 2026 consumer survey found that 58% of customers still prefer human agents for financial transactions and healthcare-related inquiries, even when they know AI can handle the interaction competently. The preference isn’t rational — it’s emotional.
The bottom line
AI in customer support isn’t an all-or-nothing proposition. The most successful deployments use AI for the high-volume, low-complexity conversations where speed and consistency matter most, while preserving human agents for the emotionally nuanced, strategically important interactions where genuine human connection creates value. The companies that get this balance right will deliver better customer experiences at lower cost. Those that automate too aggressively will save money in the short term and lose customers in the long term.