AI in CX: What’s Working, What’s Not, and What Comes Next

The best contact centers have always changed to meet the expectations of their customers, and that’s especially true as AI in CX continues to reshape how those expectations are formed. The full impact of ChatGPT’s launch in November 2022 has been difficult to quantify. The pace of artificial intelligence innovation makes it impossible to plan. What’s new today can feel obsolete tomorrow. It’s equally hard to plan when consumer behavior is caught up in the AI frenzy. 

The challenge: AI is moving faster than strategy 

So far, the rule tends to be that consumers are increasingly willing to interact with AI instead of a live agent. But that depends on the demographics of your customer base and the work your contact center performs. There are some consistent patterns. The closer an interaction comes to money, health, personal information or a problem, the more customers want a human agent. The more informational the interaction, like flight status or reservations, the higher the comfort with an AI agent. Organizations providing access to more information through self-service have seen rising satisfaction. Multiple studies support this trend,  including the Five9 Customer Experience report, which found that 72% of consumers are interested in using AI agents for quicker answers. This shift reflects a broader trend in AI in customer experience, where speed and convenience are increasingly valued, so long as they don’t come at the cost of trust.  

At the other end, consumer use of AI has exploded. I feel the use of consumer AI assistants like ChatGPT, Google’s Gemini, and Anthropic’s Claude will fundamentally change how consumers interact with brands. I see this shift as having the same impact on corporate websites that those websites had on the Yellow Pages. Anticipating this, companies are shifting from Search Engine Optimization (SEO), which aims to push a company to the top of search results, to Generative Engine Optimization (GEO). GEO structures brand data, so it’s more likely to be recommended or cited by consumer AI assistants. How does the contact center serve consumers coming from a wholly different, third-party originated experience? 

The hidden risk: losing context in AI-driven journeys 

This new avenue of consumer AI assistants “transferring” a consumer to a business’ AI or human agent is not just another digital channel for the contact center. Unlike traditional automation, autonomous AI for customer service introduces a layer of unpredictability in how consumers arrive and what context they carry with them. An AI assistant isn’t constrained by the information contained in an organizationally approved source (e.g. a website). This means a consumer being referred by AI is potentially without context or conversational history —creating friction for both the customer and the AI or human agent — and may include erroneous information or a mix of information from multiple sources (e.g. consumer wants a chocolate cake but you only sell white cake with chocolate frosting).  

There is an echo here of the days of endlessly repeating account numbers. 

There have been multiple announcements beginning in October 2025 of retailers doing just this. Early attempts at autonomous AI for customer service highlight both the potential and the complexity of executing it well. While there has been little published data on success rates to date, some high-profile pilots offer insight. Walmart was one of the first to announce integration with ChatGPT, but just six months later, in March 2026, Walmart closed the pilot, citing no increase in sales. Barely a week later, OpenAI “pivoted away” from its Instant Checkout feature to concentrate on product comparisons and recommendations through ChatGPT Ads, which are less complex and potentially easier to monetize. Meanwhile, Perplexity has been sued by Amazon to keep its consumer shopping agent off Amazon. 

Most organizations don’t have the staff or budget for such trials. For them, the use of AI needs to be focused on the use cases with the highest ROI in their vertical market from stable suppliers as part of a broader CX transformation strategy, not isolated experimentation. This seems obvious, but maybe not for the reasons you think.  

Reality check. Economics still matter. 

There is growing scrutiny around AI investment and long-term economics, often described as an “AI bubble”. Much of the current momentum is tied to continued financing and high expectations for productivity. If either shifts, it could expose challenges in the underlying business models of some AI providers.  To reach profitability, Gartner, in a December 2025 paper titled “Predicts 2026: Generative AI Will Cost a Lot More Than You Think”, believes LLM costs will rise enough that by 2030, the cost per resolution using generative AI will exceed $3, higher than some offshore human agents.  In some cases, resolution-based pricing is already approaching or exceeding that threshold today, reinforcing the need for careful evaluation of both cost and value. Meanwhile, in May 2026, GitHub announced that Microsoft Copilot would charge different rates based on the LLM used. Will exposing the specific LLM’s pricing per workload have similar results to fast food restaurants posting calorie counts? 

This problem goes well beyond the question of LLM pricing models if OpenAI, etc. are forced into profitability. It will be felt downstream as well. In the past three years, there have been hundreds of AI startups reliant on investor financing. Should they face a financing shortfall, it will be catastrophic, likened by some to the Dot-com crash where only profitable, spread-portfolio suppliers survived. The obvious operational risk from buying and building-out solutions from a small, unprofitable supplier is magnified in an investment bubble. Shiny new things are inherently fragile.  

What this means for contact center leaders 

The bottom line for contact center leaders navigating AI in CX is this: focus on the use cases that drive measurable impact, and choose sound partners that can scale with you over time. The goal isn’t to chase what’s new; it’s to build a foundation that can adapt as the market evolves as a part of a long-term CX transformation strategy.