Customer experience provider TDCX opened an AI-enabled campus in Foshan, China, in September 2024, its sixth center in the country. The facility launched with roughly 100 positions supporting Cantonese, Mandarin and regional operations, but the more consequential shift is not the technology on site - it is how the operating model redesigns frontline work around AI, from quality assurance to the way roles are structured.
In a Q&A with TNGlobal, Michael Cowell, Managing Director of TDCX Hong Kong, detailed where AI is changing CX delivery, why automation can increase the complexity of human work, and what separates an AI-native operation from one that has simply bolted tools onto existing processes.
AI's real impact is behind the conversation
The most visible AI applications - chatbots and virtual assistants - get the attention. Cowell said the bigger shift has happened where customers never look. Quality assurance is the clearest example. A traditional operation might review 2 to 5 percent of interactions. With AI-assisted review, TDCX now examines effectively every contact. "Instead of hunting for problems, they can coach against patterns the system has already surfaced," Cowell said.
Knowledge retrieval is another area. An agent handling a complex billing question no longer needs to hold the answer in their head or search a portal while the customer waits. The relevant policy surfaces in seconds. AI also reduces administrative work around each contact - summarization, disposition coding and follow-up drafting.
These changes were built into the Foshan campus from day one. Every pod uses TDCX tools across quality assurance, agent-assist workflows and voice-of-customer analytics. For professionals in AI for Customer Support, the takeaway is clear: the technology that matters most is the kind that changes what supervisors and agents do with their time, not the kind customers chat with.
What AI handles well - and what still needs a person
AI performs well when tasks are bounded and the source of truth is clear. Retrieval, summarization, translation drafts, quality review, forecasting, scheduling and high-volume transactions like order status or password resets all fall into this category.
What still requires human judgment is anything involving money, identity, emotion or ambiguity. "A customer disputing a charge is not simply looking for information," Cowell said. "They want a fair decision and someone who understands the situation." Knowing when to step outside the standard process, and when to escalate, remains human work.
There is a second-order effect. As automation absorbs simpler volume, the mix of contacts reaching a person gets harder and cognitive load rises. Operations that automate easy work but keep the same staffing and training model can see quality slip. Cowell said success requires a broader transformation plan, not just a tool. Supervisors pursuing an AI Learning Path for Call Center Supervisors will recognize this pattern: the technology changes the work before it reduces the headcount.
Language, code-switching and why Cantonese is harder
Written channels have improved significantly. Translating and maintaining a knowledge base across languages used to be a project. It is now closer to a workflow, and consistency can be better than when each market maintained its own version.
Voice is harder, and Cantonese illustrates why. Written standard Chinese and spoken Cantonese diverge significantly. Hong Kong customers often code-switch into English mid-conversation. Speech recognition handles clean, single-language audio well but still struggles with accented, code-switched, real-world speech. Register adds another layer - how directly a request can be declined differs across Japan, Hong Kong and Australia. Getting that wrong can sound rude even when every word is technically correct. TDCX built Foshan around native Cantonese and Mandarin talent rather than relying on a translation layer.
Metrics that outlast handling time
Handling time can become misleading when AI removes short, simple contacts. Average handle time can rise even as the operation becomes more efficient, because the remaining cases are more complex. TDCX built what Cowell called a fair-score mechanism: analytics teams set baseline handling time and satisfaction expectations for different contact types, then weight performance by the actual mix a person handles.
He gave more weight to repeat-contact rate within seven days, cost per resolved issue rather than cost per contact, escalation rates in both directions, and satisfaction segmented by whether AI was involved. Containment also deserves skepticism. "It tells you the customer left the automated channel, not necessarily that the problem was resolved," he said.
What makes an operation AI-native
Cowell offered a simple test: "If you switched the AI off tomorrow, what would break?" In an operation that has added tools, things get slower. In one that is genuinely AI-native, the process stops because it was designed around the AI rather than added on top of it.
The difference often shows in the organization before the technology. An AI-native operation has different staffing ratios, job descriptions and quality functions, and sometimes a different compensation model because volume handled is no longer the main measure of value. The other marker is ownership. Where AI is treated as just a tool, technology owns the deployment while operations owns the result. Where it is native, the same team owns both. Cowell said most operations are still at the bolt-on stage, but the gap should become more visible in measures like cost per resolved issue over the next few years.
Why this matters for customer support professionals
The volume of simple, scripted work will go down. The complexity of what remains will go up. A new kind of role is emerging around supervising AI itself - reviewing what it drafts, curating the knowledge it draws on, and identifying where it is wrong. Written reasoning, judgment and comfort working alongside AI tools become more important. One of the most valuable frontline capabilities is being able to look at an AI-generated answer and quickly decide whether it is correct, needs adjustment or has to go to a human. Teams will become leaner and more skilled, and the work each person does will be more valuable than the routine volume AI absorbs.
Your membership also unlocks:
